Thread: Strange Create Index behaviour
Platform: FreeBSD 6.0, Postgresql 8.1.2 compiled from the ports collection.
Not sure if this belongs in performance or bugs..
A pg_restore of my 2.5GB database was taking up to 2 hours to complete
instead of the expected 10-15 minutes. Checking the server it was mostly
CPU bound. Testing further I discovered that is was spending huge
amounts of CPU time creating some indexes.
It took a while to find out, but basically it boils down to this:
If the column that is having the index created has a certain
distribution of values then create index takes a very long time. If the
data values (integer in this case) a fairly evenly distributed then
create index is very quick, if the data values are all the same it is
very quick. I discovered that in the slow cases the column had
approximately half the values as zero and the rest fairly spread out.
One column started off with around 400,000 zeros and the rest of the
following rows spread between values of 1 to 500,000.
I have put together a test case that demonstrates the problem (see
below). I create a simple table, as close in structure to one of my
problem tables and populate an integer column with 100,000 zeros follow
by 100,000 random integers between 0 and 100,000. Then create an index
on this column. I then drop the table and repeat. The create index
should take around 1-2 seconds. A fair proportion of the time it takes
50 seconds!!!
If I fill the same row with all random data the create index always
takes a second or two. If I fill the column with all zeros everything is
still OK.
When my tables that I am trying to restore are over 2 million rows the
creating one index can take an hour!! (almost all CPU time).
All other areas of performance, once the dump is restored and analysed
seem to be OK, even large hash/merge joins and sorts
This is entirely repeatable in FreeBSD in that around half the time
create index will be incredibly slow.
All postgresql.conf settings are at the defaults for the test initially
(fresh install)
The final interesting thing is that as I increase shared buffers to 2000
or 3000 the problem gets *worse*
The following text is output from the test script..
select version();
version
------------------------------------------------------------------------------------------------
PostgreSQL 8.1.2 on i386-portbld-freebsd6.0, compiled by GCC cc (GCC)
3.4.4 [FreeBSD] 20050518
(1 row)
\timing
Timing is on.
----- Many slow cases, note the 50+ seconds cases
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 81.859 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 1482.141 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1543.508 ms
create index idx on atest(r);
CREATE INDEX
Time: 56685.230 ms
drop table atest;
DROP TABLE
Time: 4.616 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 6.889 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 2009.787 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1828.663 ms
create index idx on atest(r);
CREATE INDEX
Time: 3991.257 ms
drop table atest;
DROP TABLE
Time: 3.796 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 19.965 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 1625.059 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 2622.827 ms
create index idx on atest(r);
CREATE INDEX
Time: 1082.799 ms
drop table atest;
DROP TABLE
Time: 4.627 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 2.953 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 2068.744 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 2671.420 ms
create index idx on atest(r);
CREATE INDEX
Time: 8047.660 ms
drop table atest;
DROP TABLE
Time: 3.675 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 2.582 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 1723.987 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 2263.131 ms
create index idx on atest(r);
CREATE INDEX
Time: 50050.308 ms
drop table atest;
DROP TABLE
Time: 52.744 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 25.370 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 2052.733 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 2631.317 ms
create index idx on atest(r);
CREATE INDEX
Time: 61440.897 ms
drop table atest;
DROP TABLE
Time: 26.137 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 24.794 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),0,now(),now();
INSERT 0 100000
Time: 2851.977 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1553.046 ms
create index idx on atest(r);
CREATE INDEX
Time: 1774.920 ms
---- Fast (Normal?) cases
drop table atest;
DROP TABLE
Time: 4.422 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 2.543 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1516.246 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1407.400 ms
create index idx on atest(r);
CREATE INDEX
Time: 903.503 ms
drop table atest;
DROP TABLE
Time: 3.820 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 22.861 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1455.556 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 2037.996 ms
create index idx on atest(r);
CREATE INDEX
Time: 718.286 ms
drop table atest;
DROP TABLE
Time: 4.503 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 3.448 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1523.540 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1261.473 ms
create index idx on atest(r);
CREATE INDEX
Time: 727.707 ms
drop table atest;
DROP TABLE
Time: 3.564 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 2.897 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1447.504 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1403.525 ms
create index idx on atest(r);
CREATE INDEX
Time: 754.577 ms
drop table atest;
DROP TABLE
Time: 4.633 ms
create table atest(i int4, r int4,d1 timestamp, d2 timestamp);
CREATE TABLE
Time: 3.196 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1618.544 ms
insert into atest (i,r,d1,d2) select
generate_series(1,100000),random()*100000,now(),now();
INSERT 0 100000
Time: 1530.450 ms
create index idx on atest(r);
CREATE INDEX
Time: 802.980 ms
drop table atest;
DROP TABLE
Time: 4.707 ms
mserver#
Regards,
Gary.
Gary Doades <gpd@gpdnet.co.uk> writes:
> Platform: FreeBSD 6.0, Postgresql 8.1.2 compiled from the ports collection.
> If the column that is having the index created has a certain
> distribution of values then create index takes a very long time. If the
> data values (integer in this case) a fairly evenly distributed then
> create index is very quick, if the data values are all the same it is
> very quick. I discovered that in the slow cases the column had
> approximately half the values as zero and the rest fairly spread out.
Interesting. I tried your test script and got fairly close times
for all the cases on two different machines:
old HPUX machine: shortest 5800 msec, longest 7960 msec
new Fedora 4 machine: shortest 461 msec, longest 608 msec
(the HPUX machine was doing other stuff at the same time, so some
of its variation is probably only noise).
So what this looks like to me is a corner case that FreeBSD's qsort
fails to handle well.
You might try forcing Postgres to use our private copy of qsort, as we
do on Solaris for similar reasons. (The easy way to do this by hand
is to configure as normal, then alter the LIBOBJS setting in
src/Makefile.global to add "qsort.o", then proceed with normal build.)
However, I think that our private copy is descended from *BSD sources,
so it might have the same failure mode. It'd be worth finding out.
> The final interesting thing is that as I increase shared buffers to 2000
> or 3000 the problem gets *worse*
shared_buffers is unlikely to impact index build time noticeably in
recent PG releases. maintenance_work_mem would affect it a lot, though.
What setting were you using for that?
Can anyone else try these test cases on other platforms?
regards, tom lane
Tom Lane wrote: > Interesting. I tried your test script and got fairly close times > for all the cases on two different machines: > old HPUX machine: shortest 5800 msec, longest 7960 msec > new Fedora 4 machine: shortest 461 msec, longest 608 msec > (the HPUX machine was doing other stuff at the same time, so some > of its variation is probably only noise). > > So what this looks like to me is a corner case that FreeBSD's qsort > fails to handle well. > > You might try forcing Postgres to use our private copy of qsort, as we > do on Solaris for similar reasons. (The easy way to do this by hand > is to configure as normal, then alter the LIBOBJS setting in > src/Makefile.global to add "qsort.o", then proceed with normal build.) > However, I think that our private copy is descended from *BSD sources, > so it might have the same failure mode. It'd be worth finding out. > >> The final interesting thing is that as I increase shared buffers to 2000 >> or 3000 the problem gets *worse* > > shared_buffers is unlikely to impact index build time noticeably in > recent PG releases. maintenance_work_mem would affect it a lot, though. > What setting were you using for that? > > Can anyone else try these test cases on other platforms? > Thanks for that. I've since tried it on Windows (pg 8.1.2) and the times were all similar, around 1200ms so it might just be BSD. I'll have to wait until tomorrow to get back to my BSD box. FreeBSD ports makes it easy to install, so I'll have to figure out how to get in and change things manually. I guess the appropriate files are still left around after the ports make command finishes, so I just edit the file and make again? If it can't be fixed though I guess we may have a problem using BSD. I'm surprised this hasn't been brought up before, the case doesn't seem *that* rare. Maybe not that many using FreeBSD? I'd certainly be interested if anyone else can repro it on FreeBSD though. Regards, Gary.
Tom Lane wrote: > shared_buffers is unlikely to impact index build time noticeably in > recent PG releases. maintenance_work_mem would affect it a lot, though. > What setting were you using for that? > Also, i tried upping maintenance_work_mem to 65536 and it didn't make much difference (maybe 10% faster for the "normal" cases). Upping the shared_buffers *definitely* makes the bad cases worse though, but I agree I don't see why... Regards, Gary.
On Wed, 2006-02-15 at 20:00 +0000, Gary Doades wrote: > I have put together a test case Please enable trace_sort=on and then repeat tests and post the accompanying log file. I think this is simply the sort taking longer depending upon the data distribution, but I'd like to know for sure. Thanks, Best Regards, Simon Riggs
I wrote:
> Interesting. I tried your test script and got fairly close times
> for all the cases on two different machines:
> old HPUX machine: shortest 5800 msec, longest 7960 msec
> new Fedora 4 machine: shortest 461 msec, longest 608 msec
> So what this looks like to me is a corner case that FreeBSD's qsort
> fails to handle well.
I tried forcing PG to use src/port/qsort.c on the Fedora machine,
and lo and behold:
new Fedora 4 machine: shortest 434 msec, longest 8530 msec
So it sure looks like this script does expose a problem on BSD-derived
qsorts. Curiously, the case that's much the worst for me is the third
in the script, while the shortest time is the first case, which was slow
for Gary. So I'd venture that the *BSD code has been tweaked somewhere
along the way, in a manner that moves the problem around without really
fixing it. (Anyone want to compare the actual FreeBSD source to what
we have?)
This is pretty relevant stuff, because there was a thread recently
advocating that we stop using the platform qsort on all platforms:
http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php
It's really interesting to see a case where port/qsort is radically
worse than other qsorts ... unless we figure that out and fix it,
I think the idea of using port/qsort everywhere has just taken a
major hit.
regards, tom lane
Tom Lane wrote: > I tried forcing PG to use src/port/qsort.c on the Fedora machine, > and lo and behold: > new Fedora 4 machine: shortest 434 msec, longest 8530 msec > > So it sure looks like this script does expose a problem on BSD-derived > qsorts. Curiously, the case that's much the worst for me is the third > in the script, while the shortest time is the first case, which was slow > for Gary. So I'd venture that the *BSD code has been tweaked somewhere > along the way, in a manner that moves the problem around without really > fixing it. (Anyone want to compare the actual FreeBSD source to what > we have?) > If I run the script again, it is not always the first case that is slow, it varies from run to run, which is why I repeated it quite a few times for the test. Interestingly, if I don't delete the table after a run, but just drop and re-create the index repeatedly it stays a pretty consistent time, either repeatedly good or repeatedly bad! Regards, Gary.
Simon Riggs <simon@2ndquadrant.com> writes:
> Please enable trace_sort=on and then repeat tests and post the
> accompanying log file.
I did this on my Fedora machine with port/qsort.c, and got the results
attached. Curiously, this run has the spikes in completely different
places than the prior one did. So the random component of the test data
is affecting the results quite a lot. There seems absolutely no doubt
that we are looking at data-dependent qsort misbehavior, though. The
CPU time eaten by performsort accounts for all but about 100 msec of the
elapsed time reported on the psql side.
regards, tom lane
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.15u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/12.43u sec elapsed 12.44 sec
LOG: internal sort ended, 9861 KB used: CPU 0.01s/12.51u sec elapsed 12.52 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/0.78u sec elapsed 0.78 sec
LOG: internal sort ended, 9861 KB used: CPU 0.02s/0.85u sec elapsed 0.87 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.01s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.01s/0.96u sec elapsed 0.97 sec
LOG: internal sort ended, 9861 KB used: CPU 0.02s/1.03u sec elapsed 1.06 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/0.31u sec elapsed 0.32 sec
LOG: internal sort ended, 9861 KB used: CPU 0.02s/0.38u sec elapsed 0.40 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/7.91u sec elapsed 7.92 sec
LOG: internal sort ended, 9861 KB used: CPU 0.02s/7.99u sec elapsed 8.01 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.01s/0.13u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.01s/0.61u sec elapsed 0.63 sec
LOG: internal sort ended, 9861 KB used: CPU 0.04s/0.67u sec elapsed 0.71 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.01s/0.13u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.01s/11.52u sec elapsed 11.54 sec
LOG: internal sort ended, 9861 KB used: CPU 0.03s/11.59u sec elapsed 11.62 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/0.45u sec elapsed 0.46 sec
LOG: internal sort ended, 9861 KB used: CPU 0.02s/0.55u sec elapsed 0.57 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.00s/0.14u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.00s/0.45u sec elapsed 0.46 sec
LOG: internal sort ended, 9861 KB used: CPU 0.04s/0.54u sec elapsed 0.57 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.02s/0.12u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.02s/0.44u sec elapsed 0.46 sec
LOG: internal sort ended, 9861 KB used: CPU 0.03s/0.55u sec elapsed 0.58 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.02s/0.13u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.02s/0.44u sec elapsed 0.46 sec
LOG: internal sort ended, 9861 KB used: CPU 0.03s/0.54u sec elapsed 0.58 sec
LOG: begin index sort: unique = f, workMem = 16384, randomAccess = f
LOG: performsort starting: CPU 0.02s/0.13u sec elapsed 0.15 sec
LOG: performsort done: CPU 0.02s/0.44u sec elapsed 0.46 sec
LOG: internal sort ended, 9861 KB used: CPU 0.04s/0.54u sec elapsed 0.59 sec
Gary Doades <gpd@gpdnet.co.uk> writes:
> Interestingly, if I don't delete the table after a run, but just drop
> and re-create the index repeatedly it stays a pretty consistent time,
> either repeatedly good or repeatedly bad!
This is consistent with the theory of a data-dependent performance
problem in qsort. If you don't generate a fresh set of random test
data, then you get repeatable runtimes. With a new set of test data,
you might or might not hit the not-so-sweet-spot that we seem to have
detected.
regards, tom lane
Tom Lane wrote: > > So it sure looks like this script does expose a problem on BSD-derived > qsorts. Curiously, the case that's much the worst for me is the third > in the script, while the shortest time is the first case, which was slow > for Gary. So I'd venture that the *BSD code has been tweaked somewhere > along the way, in a manner that moves the problem around without really > fixing it. (Anyone want to compare the actual FreeBSD source to what > we have?) > > It's really interesting to see a case where port/qsort is radically > worse than other qsorts ... unless we figure that out and fix it, > I think the idea of using port/qsort everywhere has just taken a > major hit. > More specifically to BSD, is there any way I can use a non-BSD qsort for building Postresql server? Regards, Gary.
Gary Doades <gpd@gpdnet.co.uk> writes: > If I run the script again, it is not always the first case that is slow, > it varies from run to run, which is why I repeated it quite a few times > for the test. For some reason I hadn't immediately twigged to the fact that your test script is just N repetitions of the exact same structure with random data. So it's not so surprising that you get random variations in behavior with different test data sets. I did some experimentation comparing the qsort from Fedora Core 4 (glibc-2.3.5-10.3) with our src/port/qsort.c. For those who weren't following the pgsql-performance thread, the test case is just this repeated a lot of times: create table atest(i int4, r int4); insert into atest (i,r) select generate_series(1,100000), 0; insert into atest (i,r) select generate_series(1,100000), random()*100000; \timing create index idx on atest(r); \timing drop table atest; I did this 100 times and sorted the reported runtimes. (Investigation with trace_sort = on confirms that the runtime is almost entirely spent in qsort() called from our performsort --- the Postgres overhead is about 100msec on this machine.) Results are below. It seems clear that our qsort.c is doing a pretty awful job of picking qsort pivots, while glibc is mostly managing not to make that mistake. I haven't looked at the glibc code yet to see what they are doing differently. I'd say this puts a considerable damper on my enthusiasm for using our qsort all the time, as was recently debated in this thread: http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php We need to fix our qsort.c before pushing ahead with that idea. regards, tom lane 100 runtimes for glibc qsort, sorted ascending: Time: 459.860 ms Time: 460.209 ms Time: 460.704 ms Time: 461.317 ms Time: 461.538 ms Time: 461.652 ms Time: 461.988 ms Time: 462.573 ms Time: 462.638 ms Time: 462.716 ms Time: 462.917 ms Time: 463.219 ms Time: 463.455 ms Time: 463.650 ms Time: 463.723 ms Time: 463.737 ms Time: 463.750 ms Time: 463.852 ms Time: 463.964 ms Time: 463.988 ms Time: 464.003 ms Time: 464.135 ms Time: 464.372 ms Time: 464.458 ms Time: 464.496 ms Time: 464.551 ms Time: 464.599 ms Time: 464.655 ms Time: 464.656 ms Time: 464.722 ms Time: 464.814 ms Time: 464.827 ms Time: 464.878 ms Time: 464.899 ms Time: 464.905 ms Time: 464.987 ms Time: 465.055 ms Time: 465.138 ms Time: 465.159 ms Time: 465.194 ms Time: 465.310 ms Time: 465.316 ms Time: 465.375 ms Time: 465.450 ms Time: 465.535 ms Time: 465.595 ms Time: 465.680 ms Time: 465.769 ms Time: 465.865 ms Time: 465.892 ms Time: 465.903 ms Time: 466.003 ms Time: 466.154 ms Time: 466.164 ms Time: 466.203 ms Time: 466.305 ms Time: 466.344 ms Time: 466.364 ms Time: 466.388 ms Time: 466.502 ms Time: 466.593 ms Time: 466.725 ms Time: 466.794 ms Time: 466.798 ms Time: 466.904 ms Time: 466.971 ms Time: 466.997 ms Time: 467.122 ms Time: 467.146 ms Time: 467.221 ms Time: 467.224 ms Time: 467.244 ms Time: 467.277 ms Time: 467.587 ms Time: 468.142 ms Time: 468.207 ms Time: 468.237 ms Time: 468.471 ms Time: 468.663 ms Time: 468.700 ms Time: 469.235 ms Time: 469.840 ms Time: 470.472 ms Time: 471.140 ms Time: 472.811 ms Time: 472.959 ms Time: 474.858 ms Time: 477.210 ms Time: 479.571 ms Time: 479.671 ms Time: 482.797 ms Time: 488.852 ms Time: 514.639 ms Time: 529.287 ms Time: 612.185 ms Time: 660.748 ms Time: 742.227 ms Time: 866.814 ms Time: 1234.848 ms Time: 1267.398 ms 100 runtimes for port/qsort.c, sorted ascending: Time: 418.905 ms Time: 420.611 ms Time: 420.764 ms Time: 420.904 ms Time: 421.706 ms Time: 422.466 ms Time: 422.627 ms Time: 423.189 ms Time: 423.302 ms Time: 425.096 ms Time: 425.731 ms Time: 425.851 ms Time: 427.253 ms Time: 430.113 ms Time: 432.756 ms Time: 432.963 ms Time: 440.502 ms Time: 440.640 ms Time: 450.452 ms Time: 458.143 ms Time: 459.212 ms Time: 467.706 ms Time: 468.006 ms Time: 468.574 ms Time: 470.003 ms Time: 472.313 ms Time: 483.622 ms Time: 492.395 ms Time: 509.564 ms Time: 531.037 ms Time: 533.366 ms Time: 535.610 ms Time: 575.523 ms Time: 582.688 ms Time: 593.545 ms Time: 647.364 ms Time: 660.612 ms Time: 677.312 ms Time: 680.288 ms Time: 697.626 ms Time: 833.066 ms Time: 834.511 ms Time: 851.819 ms Time: 920.443 ms Time: 926.731 ms Time: 954.289 ms Time: 1045.214 ms Time: 1059.200 ms Time: 1062.328 ms Time: 1136.018 ms Time: 1260.091 ms Time: 1276.883 ms Time: 1319.351 ms Time: 1438.854 ms Time: 1475.457 ms Time: 1538.211 ms Time: 1549.004 ms Time: 1744.642 ms Time: 1771.258 ms Time: 1959.530 ms Time: 2300.140 ms Time: 2589.641 ms Time: 2612.780 ms Time: 3100.024 ms Time: 3284.125 ms Time: 3379.792 ms Time: 3750.278 ms Time: 4302.278 ms Time: 4780.624 ms Time: 5000.056 ms Time: 5092.604 ms Time: 5168.722 ms Time: 5292.941 ms Time: 5895.964 ms Time: 7003.164 ms Time: 7099.449 ms Time: 7115.083 ms Time: 7384.940 ms Time: 8214.010 ms Time: 8700.771 ms Time: 9331.225 ms Time: 10503.360 ms Time: 12496.026 ms Time: 12982.474 ms Time: 15192.390 ms Time: 15392.161 ms Time: 15958.295 ms Time: 18375.693 ms Time: 18617.706 ms Time: 18927.515 ms Time: 19898.018 ms Time: 20865.979 ms Time: 21000.907 ms Time: 21297.585 ms Time: 21714.518 ms Time: 25423.235 ms Time: 27543.052 ms Time: 28314.182 ms Time: 29400.278 ms Time: 34142.534 ms
On Wed, 2006-02-15 at 16:51 -0500, Tom Lane wrote: > Gary Doades <gpd@gpdnet.co.uk> writes: > > Interestingly, if I don't delete the table after a run, but just drop > > and re-create the index repeatedly it stays a pretty consistent time, > > either repeatedly good or repeatedly bad! > > This is consistent with the theory of a data-dependent performance > problem in qsort. If you don't generate a fresh set of random test > data, then you get repeatable runtimes. With a new set of test data, > you might or might not hit the not-so-sweet-spot that we seem to have > detected. Agreed. Good analysis... Best Regards, Simon Riggs
Tom Lane wrote: > For some reason I hadn't immediately twigged to the fact that your test > script is just N repetitions of the exact same structure with random data. > So it's not so surprising that you get random variations in behavior > with different test data sets. > > It seems clear that our qsort.c is doing a pretty awful job of picking > qsort pivots, while glibc is mostly managing not to make that mistake. > I haven't looked at the glibc code yet to see what they are doing > differently. > > I'd say this puts a considerable damper on my enthusiasm for using our > qsort all the time, as was recently debated in this thread: > http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php > We need to fix our qsort.c before pushing ahead with that idea. [snip] > Time: 28314.182 ms > Time: 29400.278 ms > Time: 34142.534 ms Ouch! That confirms my problem. I generated the random test case because it was easier than including the dump of my tables, but you can appreciate that tables 20 times the size are basically crippled when it comes to creating an index on them. Examining the dump and the associated times during restore it looks like I have 7 tables with this approximate distribution, thus the ridiculously long restore time. Better not re-index soon! Is this likely to hit me in a random fashion during normal operation, joins, sorts, order by for example? So the options are: 1) Fix the included qsort.c code and use that 2) Get FreeBSD to fix their qsort code 3) Both I guess that 1 is the real solution in case anyone else's qsort is broken in the same way. Then at least you *could* use it all the time :) Regards, Gary.
Gary Doades <gpd@gpdnet.co.uk> writes:
> Is this likely to hit me in a random fashion during normal operation,
> joins, sorts, order by for example?
Yup, anytime you're passing data with that kind of distribution
through a sort.
> So the options are:
> 1) Fix the included qsort.c code and use that
> 2) Get FreeBSD to fix their qsort code
> 3) Both
> I guess that 1 is the real solution in case anyone else's qsort is
> broken in the same way. Then at least you *could* use it all the time :)
It's reasonable to assume that most of the *BSDen have basically the
same qsort code. Ours claims to have come from NetBSD sources, but
I don't doubt that they all trace back to a common ancestor.
regards, tom lane
Gary Doades <gpd@gpdnet.co.uk> writes:
> Ouch! That confirms my problem. I generated the random test case because
> it was easier than including the dump of my tables, but you can
> appreciate that tables 20 times the size are basically crippled when it
> comes to creating an index on them.
Actually... we only use qsort when we have a sorting problem that fits
within the allowed sort memory. The external-sort logic doesn't go
through that code at all. So all the analysis we just did on your test
case doesn't necessarily apply to sort problems that are too large for
the sort_mem setting.
The test case would be sorting 200000 index entries, which'd probably
occupy at least 24 bytes apiece of sort memory, so probably about 5 meg.
A problem 20 times that size would definitely not fit in the default
16MB maintenance_work_mem. Were you using a large value of
maintenance_work_mem for your restore?
regards, tom lane
This behavior is consistent with the pivot choosing algorithm assuming certain distribution(s) for the data. For instance, median-of-three partitioning is known to be pessimal when the data is geometrically or hyper-geometrically distributed. Also, care must be taken that sometimes is not when there are many equal values in the data. Even pseudo random number generator based pivot choosing algorithms are not immune if the PRNG is flawed in some way. How are we choosing our pivots? At 06:28 PM 2/15/2006, Tom Lane wrote: >I did some experimentation comparing the qsort from Fedora Core 4 >(glibc-2.3.5-10.3) with our src/port/qsort.c. For those who weren't >following the pgsql-performance thread, the test case is just this >repeated a lot of times: > >create table atest(i int4, r int4); >insert into atest (i,r) select generate_series(1,100000), 0; >insert into atest (i,r) select generate_series(1,100000), random()*100000; >\timing >create index idx on atest(r); >\timing >drop table atest; > >I did this 100 times and sorted the reported runtimes. (Investigation >with trace_sort = on confirms that the runtime is almost entirely spent >in qsort() called from our performsort --- the Postgres overhead is >about 100msec on this machine.) Results are below. > >It seems clear that our qsort.c is doing a pretty awful job of picking >qsort pivots, while glibc is mostly managing not to make that mistake. >I haven't looked at the glibc code yet to see what they are doing >differently. > >I'd say this puts a considerable damper on my enthusiasm for using our >qsort all the time, as was recently debated in this thread: >http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php >We need to fix our qsort.c before pushing ahead with that idea. > > regards, tom lane > > >100 runtimes for glibc qsort, sorted ascending: > >Time: 459.860 ms ><snip> >Time: 488.852 ms >Time: 514.639 ms >Time: 529.287 ms >Time: 612.185 ms >Time: 660.748 ms >Time: 742.227 ms >Time: 866.814 ms >Time: 1234.848 ms >Time: 1267.398 ms > > >100 runtimes for port/qsort.c, sorted ascending: > >Time: 418.905 ms ><snip> >Time: 20865.979 ms >Time: 21000.907 ms >Time: 21297.585 ms >Time: 21714.518 ms >Time: 25423.235 ms >Time: 27543.052 ms >Time: 28314.182 ms >Time: 29400.278 ms >Time: 34142.534 ms
I wrote:
> Gary Doades <gpd@gpdnet.co.uk> writes:
>> Ouch! That confirms my problem. I generated the random test case because
>> it was easier than including the dump of my tables, but you can
>> appreciate that tables 20 times the size are basically crippled when it
>> comes to creating an index on them.
> Actually... we only use qsort when we have a sorting problem that fits
> within the allowed sort memory. The external-sort logic doesn't go
> through that code at all. So all the analysis we just did on your test
> case doesn't necessarily apply to sort problems that are too large for
> the sort_mem setting.
I increased the size of the test case by 10x (basically s/100000/1000000/)
which is enough to push it into the external-sort regime. I get
amazingly stable runtimes now --- I didn't have the patience to run 100
trials, but in 30 trials I have slowest 11538 msec and fastest 11144 msec.
So this code path is definitely not very sensitive to this data
distribution.
While these numbers aren't glittering in comparison to the best-case
qsort times (~450 msec to sort 10% as much data), they are sure a lot
better than the worst-case times. So maybe a workaround for you is
to decrease maintenance_work_mem, counterintuitive though that be.
(Now, if you *weren't* using maintenance_work_mem of 100MB or more
for your problem restore, then I'm not sure I know what's going on...)
We still ought to try to fix qsort of course.
regards, tom lane
Ron <rjpeace@earthlink.net> writes:
> How are we choosing our pivots?
See qsort.c: it looks like median of nine equally spaced inputs (ie,
the 1/8th points of the initial input array, plus the end points),
implemented as two rounds of median-of-three choices. With half of the
data inputs zero, it's not too improbable for two out of the three
samples to be zeroes in which case I think the med3 result will be zero
--- so choosing a pivot of zero is much more probable than one would
like, and doing so in many levels of recursion causes the problem.
I think. I'm not too sure if the code isn't just being sloppy about the
case where many data values are equal to the pivot --- there's a special
case there to switch to insertion sort, and maybe that's getting invoked
too soon. It'd be useful to get a line-level profile of the behavior of
this code in the slow cases...
regards, tom lane
> Ouch! That confirms my problem. I generated the random test case because > it was easier than including the dump of my tables, but you can > appreciate that tables 20 times the size are basically crippled when it > comes to creating an index on them. I have to say that I restored a few gigabyte dump on freebsd the other day, and most of the restore time was in index creation - I didn't think too much of it though at the time. FreeBSD 4.x. Chris
On Wed, 2006-02-15 at 20:00 +0000, Gary Doades wrote: > I have put together a test case that demonstrates the problem (see > below). I create a simple table, as close in structure to one of my > problem tables and populate an integer column with 100,000 zeros follow > by 100,000 random integers between 0 and 100,000. Then create an index > on this column. I then drop the table and repeat. The create index > should take around 1-2 seconds. A fair proportion of the time it takes > 50 seconds!!! > > If I fill the same row with all random data the create index always > takes a second or two. If I fill the column with all zeros everything is > still OK. Aside from the importance of investigating sort behaviour, have you tried to build a partial index WHERE col > 0 ? That way you wouldn't even be indexing the zeros. Best Regards, Simon Riggs
On Wed, 2006-02-15 at 19:59 -0500, Tom Lane wrote: > I get > amazingly stable runtimes now --- I didn't have the patience to run 100 > trials, but in 30 trials I have slowest 11538 msec and fastest 11144 msec. > So this code path is definitely not very sensitive to this data > distribution. "The worst-case behavior of replacement-selection is very close to its average behavior, while the worst-case behavior of QuickSort is terrible (N2) – a strong argument in favor of replacement-selection. Despite this risk, QuickSort is widely used because, in practice, it has superior performance." p.8, "AlphaSort: A Cache-Sensitive Parallel External Sort", Nyberg et al, VLDB Journal 4(4): 603-627 (1995) I think your other comment about flipping to insertion sort too early (and not returning...) is a plausible cause for the poor pg qsort behaviour, but the overall spread of values seems as expected. Some test results I've seen seem consistent with the view that increasing memory also increases run-time for larger settings of work_mem/maintenance_work_mem. Certainly, as I observed a while back, having a large memory settings doesn't help you at all when you are doing final run merging on the external sort. Whatever we do, we should look at the value high memory settings bring to each phase of a sort separately from the other phases. There is work underway on improving external sorts, so I hear (not me). Plus my WIP on randomAccess requirements. Best Regards, Simon Riggs
On Wed, 2006-02-15 at 18:28 -0500, Tom Lane wrote: > It seems clear that our qsort.c is doing a pretty awful job of picking > qsort pivots, while glibc is mostly managing not to make that mistake. > I haven't looked at the glibc code yet to see what they are doing > differently. glibc qsort is actually merge sort, so I'm not surprised it avoids this problem. -Neil
At 08:21 PM 2/15/2006, Tom Lane wrote:
>Ron <rjpeace@earthlink.net> writes:
> > How are we choosing our pivots?
>
>See qsort.c: it looks like median of nine equally spaced inputs (ie,
>the 1/8th points of the initial input array, plus the end points),
>implemented as two rounds of median-of-three choices.
OK, this is a bad way to do median-of-n partitioning for a few
reasons. See Sedgewick's PhD thesis for details.
Basically, if one is using median-of-n partitioning to choose a
pivot, one should do it in =one= pass, and n for that pass should be
<= the numbers of registers in the CPU. Since the x86 ISA has 8
GPR's, n should be <= 8. 7 for instance.
Special purposing the median-of-n code so that the minimal number of
comparisons and moves is used to sort the sample and then
"partitioning in place" is the best way to do it. In addition, care
must be taken to deal with the possibility that many of the keys may be equal.
The (pseudo) code looks something like this:
qs(a[],L,R){
if((R-L) > SAMPLE_SIZE){ // Not worth using qs for too few elements
SortSample(SAMPLE_SIZE,a[],L,R);
// Sorts SAMPLE_SIZE= n elements and does median-of-n
partitioning for small n
// using the minimal number of comparisons and moves.
// In the process it ends up partitioning the first n/2 and last
n/2 elements
// SAMPLE_SIZE is a constant chosen to work best for a given CPU.
// #GPRs - 1 is a good initial guess.
// For the x86 ISA, #GPRs - 1 = 7. For native x86-64, it's 15.
// For most RISC CPUs it's 31 or 63. For Itanium, it's 127 (!)
pivot= a[(L+R)>>1]; i= L+(SAMPLE_SIZE>>1); j= R-(SAMPLE_SIZE>>1);
for(;;){
while(a[++i] < pivot);
while(a[--j] > pivot);
if(i >= j) break;
if(a[i] > a[j]) swap(a[i],a[j]);
}
if((i-R) >= (j-L)){qs(a,L,i-1);}
else{qs(a,i,R);}
else{OofN^2_Sort(a,L,R);}
// SelectSort may be better than InsertSort if KeySize in bits <<
RecordSize in bits
} // End of qs
Given that the most common CPU ISA in existence has 8 GPRs,
SAMPLE_SIZE= 7 is probably optimal:
t= (L+R);
the set would be {L; t/8; t/4; t/2; 3*t/4; 7*t/8; R;}
==> {L; t>>3; t>>2; t>>1; (3*t)>>2; (7*t)>>3; R} as the locations.
Even better (and more easily scaled as the number of GPR's in the CPU
changes) is to use
the set {L; L+1; L+2; t>>1; R-2; R-1; R}
This means that instead of 7 random memory accesses, we have 3; two
of which result in a
burst access for three elements each.
That's much faster; _and_ using a sample of 9, 15, 31, 63, etc (to
max of ~GPRs -1) elements is more easily done.
It also means that the work we do sorting the sample can be taken
advantage of when starting
inner loop of quicksort: items L..L+2, t, and R-2..R are already
partitioned by SortSample().
Insuring that the minimum number of comparisons and moves is done in
SortSample can be down by using a code generator to create a
comparison tree that identifies which permutation(s) of n we are
dealing with and then moving them into place with the minimal number of moves.
SIDE NOTE: IIRC glibc's qsort is actually merge sort. Merge sort
performance is insensitive to all inputs, and there are way to
optimize it as well.
I'll leave the actual coding to someone who knows the pg source
better than I do.
Ron
Tom Lane wrote: > I increased the size of the test case by 10x (basically s/100000/1000000/) > which is enough to push it into the external-sort regime. I get > amazingly stable runtimes now --- I didn't have the patience to run 100 > trials, but in 30 trials I have slowest 11538 msec and fastest 11144 msec. > So this code path is definitely not very sensitive to this data > distribution. > > While these numbers aren't glittering in comparison to the best-case > qsort times (~450 msec to sort 10% as much data), they are sure a lot > better than the worst-case times. So maybe a workaround for you is > to decrease maintenance_work_mem, counterintuitive though that be. > (Now, if you *weren't* using maintenance_work_mem of 100MB or more > for your problem restore, then I'm not sure I know what's going on...) > Good call. I basically reversed your test by keeping the number of rows the same (200000), but reducing maintenance_work_mem. Reducing to 8192 made no real difference. Reducing to 4096 flattened out all the times nicely. Slower overall, but at least predictable. Hopefully only a temporary solution until qsort is fixed. My restore now takes 22 minutes :) I think the reason I wasn't seeing performance issues with normal sort operations is because they use work_mem not maintenance_work_mem which was only set to 2048 anyway. Does that sound right? Regards, Gary.
On Wed, Feb 15, 2006 at 11:30:54PM -0500, Ron wrote:
> Even better (and more easily scaled as the number of GPR's in the CPU
> changes) is to use
> the set {L; L+1; L+2; t>>1; R-2; R-1; R}
> This means that instead of 7 random memory accesses, we have 3; two
> of which result in a
> burst access for three elements each.
Isn't that improvement going to disappear competely if you choose a bad
pivot?
> SIDE NOTE: IIRC glibc's qsort is actually merge sort. Merge sort
> performance is insensitive to all inputs, and there are way to
> optimize it as well.
glibc-2.3.5/stdlib/qsort.c:
/* Order size using quicksort. This implementation incorporates
four optimizations discussed in Sedgewick:
I can't see any references to merge sort in there at all.
/* Steinar */
--
Homepage: http://www.sesse.net/
* Neil Conway: > On Wed, 2006-02-15 at 18:28 -0500, Tom Lane wrote: >> It seems clear that our qsort.c is doing a pretty awful job of picking >> qsort pivots, while glibc is mostly managing not to make that mistake. >> I haven't looked at the glibc code yet to see what they are doing >> differently. > > glibc qsort is actually merge sort, so I'm not surprised it avoids this > problem. qsort also performs twice as many key comparisons as the theoretical minimum. If key comparison is not very cheap, other schemes (like heapsort, for example) are more attractive.
On Thu, Feb 16, 2006 at 01:10:48PM +0100, Florian Weimer wrote: > * Neil Conway: > > > On Wed, 2006-02-15 at 18:28 -0500, Tom Lane wrote: > >> It seems clear that our qsort.c is doing a pretty awful job of picking > >> qsort pivots, while glibc is mostly managing not to make that mistake. > >> I haven't looked at the glibc code yet to see what they are doing > >> differently. > > > > glibc qsort is actually merge sort, so I'm not surprised it avoids this > > problem. > > qsort also performs twice as many key comparisons as the theoretical > minimum. If key comparison is not very cheap, other schemes (like > heapsort, for example) are more attractive. Last time around there were a number of different algorithms tested. Did anyone run those tests while getting it to count the number of actual comparisons (which could easily swamp the time taken to do the actual sort in some cases)? Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
Attachment
Martijn van Oosterhout schrieb: > > Last time around there were a number of different algorithms tested. > Did anyone run those tests while getting it to count the number of > actual comparisons (which could easily swamp the time taken to do the > actual sort in some cases)? > The last time I did such tests is almost 10 years ago. I had used MetroWerks CodeWarrior C/C++, which had Quicksort as algorithm in the Lib C. Anyhow, I tested a few algorithms including merge sort and heapsort. I end up with heapsort because it was the fastest algorithm for our issue. We joined two arrays where each array was sorted and run qsort to sort the new array. Sven.
At 06:35 AM 2/16/2006, Steinar H. Gunderson wrote:
>On Wed, Feb 15, 2006 at 11:30:54PM -0500, Ron wrote:
> > Even better (and more easily scaled as the number of GPR's in the CPU
> > changes) is to use
> > the set {L; L+1; L+2; t>>1; R-2; R-1; R}
> > This means that instead of 7 random memory accesses, we have 3; two
> > of which result in a burst access for three elements each.
>
>Isn't that improvement going to disappear competely if you choose a bad
>pivot?
Only if you _consistently_ (read: "the vast majority of the time":
quicksort is actually darn robust) choose a _pessimal_, not just
"bad", pivot quicksort will degenerate to the O(N^2) behavior
everyone worries about. See Corman & Rivest for a proof on this.
Even then, doing things as above has benefits:
1= The worst case is less bad since the guaranteed O(lgs!) pivot
choosing algorithm puts s elements into final position.
Worst case becomes better than O(N^2/(s-1)).
2= The overhead of pivot choosing can overshadow the benefits using
more traditional methods for even moderate values of s. See
discussions on the quicksort variant known as "samplesort" and
Sedgewick's PhD thesis for details. Using a pivot choosing algorithm
that actually does some of the partitioning (and does it more
efficiently than the "usual" partitioning algorithm does) plus using
partition-in-place (rather then Lomuto's method) reduces overhead
very effectively (at the "cost" of more complicated / delicate to get
right partitioning code). The above reduces the number of moves used
in a quicksort pass considerably regardless of the number of compares used.
3= Especially in modern systems where the gap between internal CPU
bandwidth and memory bandwidth is so great, the overhead of memory
accesses for comparisons and moves is the majority of the overhead
for both the pivot choosing and the partitioning algorithms within
quicksort. Particularly random memory accesses. The reason (#GPRs -
1) is a magic constant is that it's the most you can compare and move
using only register-to-register operations.
In addition, replacing as many of the memory accesses you must do
with sequential rather than random memory accesses is a big deal:
sequential memory access is measured in 10's of CPU cycles while
random memory access is measured in hundreds of CPU cycles. It's no
accident that the advances in Grey's sorting contest have involved
algorithms that are both register and cache friendly, minimizing
overall memory access and using sequential memory access as much as
possible when said access can not be avoided. As caches grow larger
and memory accesses more expensive, it's often worth it to use a
BucketSort+QuickSort hybrid rather than just QuickSort.
...and of course if you know enough about the data to be sorted so as
to constrain it appropriately, one should use a non comparison based
O(N) sorting algorithm rather than any of the general comparison
based O(NlgN) methods.
> > SIDE NOTE: IIRC glibc's qsort is actually merge sort. Merge sort
> > performance is insensitive to all inputs, and there are way to
> > optimize it as well.
>
>glibc-2.3.5/stdlib/qsort.c:
>
> /* Order size using quicksort. This implementation incorporates
> four optimizations discussed in Sedgewick:
>
>I can't see any references to merge sort in there at all.
Well, then I'm not the only person on the lists whose memory is faulty ;-)
The up side of MergeSort is that its performance is always O(NlgN).
The down sides are that it is far more memory hungry than QuickSort and slower.
Ron
At 07:10 AM 2/16/2006, Florian Weimer wrote: >* Neil Conway: > > > On Wed, 2006-02-15 at 18:28 -0500, Tom Lane wrote: > >> It seems clear that our qsort.c is doing a pretty awful job of picking > >> qsort pivots, while glibc is mostly managing not to make that mistake. > >> I haven't looked at the glibc code yet to see what they are doing > >> differently. > > > > glibc qsort is actually merge sort, so I'm not surprised it avoids this > > problem. > >qsort also performs twice as many key comparisons as the theoretical >minimum. The theoretical minimum number of comparisons for a general purpose comparison based sort is O(lgN!). QuickSort uses 2NlnN ~= 1.38NlgN or ~1.38x the optimum without tuning (see Knuth, Sedgewick, Corman, ... etc) OTOH, QuickSort uses ~2x as many =moves= as the theoretical minimum unless tuned, and moves are more expensive than compares in modern systems. See my other posts for QuickSort tuning methods that attempt to directly address both issues. Ron
Hi, Ron, Ron wrote: > ...and of course if you know enough about the data to be sorted so as to > constrain it appropriately, one should use a non comparison based O(N) > sorting algorithm rather than any of the general comparison based > O(NlgN) methods. Sounds interesting, could you give us some pointers (names, URLs, papers) to such algorithms? Thanks a lot, Markus -- Markus Schaber | Logical Tracking&Tracing International AG Dipl. Inf. | Software Development GIS Fight against software patents in EU! www.ffii.org www.nosoftwarepatents.org
"Gary Doades" <gpd@gpdnet.co.uk> writes:
> I think the reason I wasn't seeing performance issues with normal sort
> operations is because they use work_mem not maintenance_work_mem which was
> only set to 2048 anyway. Does that sound right?
Very probable. Do you want to test the theory by jacking that up? ;-)
regards, tom lane
On Thu, Feb 16, 2006 at 08:22:55AM -0500, Ron wrote: > 3= Especially in modern systems where the gap between internal CPU > bandwidth and memory bandwidth is so great, the overhead of memory > accesses for comparisons and moves is the majority of the overhead > for both the pivot choosing and the partitioning algorithms within > quicksort. Particularly random memory accesses. The reason (#GPRs - > 1) is a magic constant is that it's the most you can compare and move > using only register-to-register operations. But how much of this applies to us? We're not sorting arrays of integers, we're sorting pointers to tuples. So while moves cost very little, a comparison costs hundreds, maybe thousands of cycles. A tuple can easily be two or three cachelines and you're probably going to access all of it, not to mention the Fmgr structures and the Datums themselves. None of this is cache friendly. The actual tuples themselves could be spread all over memory (I don't think any particular effort is expended trying to minimize fragmentation). Do these algorithms discuss the case where a comparison is more than 1000 times the cost of a move? Where this does become interesting is where we can convert a datum to an integer such that if f(A) > f(B) then A > B. Then we can sort on f(X) first with just integer comparisons and then do a full tuple comparison only if f(A) = f(B). This would be much more cache-coherent and make these algorithms much more applicable in my mind. Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
Attachment
> "Gary Doades" <gpd@gpdnet.co.uk> writes: >> I think the reason I wasn't seeing performance issues with normal sort >> operations is because they use work_mem not maintenance_work_mem which >> was >> only set to 2048 anyway. Does that sound right? > > Very probable. Do you want to test the theory by jacking that up? ;-) Hmm, played around a bit. I have managed to get it to do a sort on one of the "bad" columns using a select of two whole tables that results in a sequntial scan, sort and merge join. I also tried a simple select column order by column for a bad column. I tried varying maintenance_work_mem and work_mem up and down between 2048 and 65536 but I always get similar results. The sort phase always takes 4 to 5 seconds which seems about right for 900,000 rows. This was on a colunm that took 12 minutes to create an index on. I've no idea why it should behave this way, but probably explains why I (and others) may not have noticed it before. Regards, Gary.
At 09:48 AM 2/16/2006, Martijn van Oosterhout wrote: >On Thu, Feb 16, 2006 at 08:22:55AM -0500, Ron wrote: > > 3= Especially in modern systems where the gap between internal CPU > > bandwidth and memory bandwidth is so great, the overhead of memory > > accesses for comparisons and moves is the majority of the overhead > > for both the pivot choosing and the partitioning algorithms within > > quicksort. Particularly random memory accesses. The reason (#GPRs - > > 1) is a magic constant is that it's the most you can compare and move > > using only register-to-register operations. > >But how much of this applies to us? We're not sorting arrays of >integers, we're sorting pointers to tuples. So while moves cost very >little, a comparison costs hundreds, maybe thousands of cycles. A tuple >can easily be two or three cachelines and you're probably going to >access all of it, not to mention the Fmgr structures and the Datums >themselves. Pointers are simply fixed size 32b or 64b quantities. They are essentially integers. Comparing and moving pointers or fixed size keys to those pointers is exactly the same problem as comparing and moving integers. Comparing =or= moving the actual data structures is a much more expensive and variable cost proposition. I'm sure that pg's sort functionality is written intelligently enough that the only real data moves are done in a final pass after the exact desired order has been found using pointer compares and (re)assignments during the sorting process. That's a standard technique for sorting data whose "key" or pointer is much smaller than a datum. Your cost comment basically agrees with mine regarding the cost of random memory accesses. The good news is that the number of datums to be examined during the pivot choosing process is small enough that the datums can fit into CPU cache while the pointers to them can be assigned to registers: making pivot choosing +very+ fast when done correctly. As you've noted, actual partitioning is going to be more expensive since it involves accessing enough actual datums that they can't all fit into CPU cache. The good news is that QuickSort has a very sequential access pattern within its inner loop. So while we must go to memory for compares, we are at least keeping the cost for it down it a minimum. In addition, said access is nice enough to be very prefetch and CPU cache hierarchy friendly. >None of this is cache friendly. The actual tuples themselves could be >spread all over memory (I don't think any particular effort is expended >trying to minimize fragmentation). It probably would be worth it to spend some effort on memory layout just as we do for HD layout. >Do these algorithms discuss the case where a comparison is more than >1000 times the cost of a move? A move is always more expensive than a compare when the datum is larger than its pointer or key. A move is always more expensive than a compare when it involves memory to memory movement rather than CPU location to CPU location movement. A move is especially more expensive than a compare when it involves both factors. Most moves do involve both. What I suspect you meant is that a key comparison that involves accessing the data in memory is more expensive than reassigning the pointers associated with those keys. That is certainly true. Yes. The problem has been extensively studied. ;-) >Where this does become interesting is where we can convert a datum to >an integer such that if f(A) > f(B) then A > B. Then we can sort on >f(X) first with just integer comparisons and then do a full tuple >comparison only if f(A) = f(B). This would be much more cache-coherent >and make these algorithms much more applicable in my mind. In fact we can do better. Using hash codes or what-not to map datums to keys and then sorting just the keys and the pointers to those datums followed by an optional final pass where we do the actual data movement is also a standard technique for handling large data structures. Regardless of what tweaks beyond the basic algorithms we use, the algorithms themselves have been well studied and their performance well established. QuickSort is the best performing of the O(nlgn) comparison based sorts and it uses less resources than HeapSort or MergeSort. Ron
Ron <rjpeace@earthlink.net> writes:
> Your cost comment basically agrees with mine regarding the cost of
> random memory accesses. The good news is that the number of datums
> to be examined during the pivot choosing process is small enough that
> the datums can fit into CPU cache while the pointers to them can be
> assigned to registers: making pivot choosing +very+ fast when done correctly.
This is more or less irrelevant given that comparing the pointers is not
the operation we need to do.
regards, tom lane
Markus Schaber wrote: > Ron wrote: >>...and of course if you know enough about the data to be sorted so as to >>constrain it appropriately, one should use a non comparison based O(N) >>sorting algorithm rather than any of the general comparison based >>O(NlgN) methods. > > Sounds interesting, could you give us some pointers (names, URLs, > papers) to such algorithms? Most of these techniques boil down to good ol' "bucket sort". A simple example: suppose you have a column of integer percentages,range zero to 100. You know there are only 101 distinct values. So create 101 "buckets" (e.g. linked lists),make a single pass through your data and drop each row's ID into the right bucket, then make a second pass throughthe buckets, and write the row ID's out in bucket order. This is an O(N) sort technique. Any time you have a restricted data range, you can do this. Say you have 100 million rows of scientific results known tobe good to only three digits -- it can have at most 1,000 distinct values (regardless of the magnitude of the values),so you can do this with 1,000 buckets and just two passes through the data. You can also use this trick when the optimizer is asked for "fastest first result." Say you have a cursor on a column ofnumbers with good distribution. If you do a bucket sort on the first two or three digits only, you know the first "page"of results will be in the first bucket. So you only need to apply qsort to that first bucket (which is very fast,since it's small), and you can deliver the first page of data to the application. This can be particularly effectivein interactive situations, where the user typically looks at a few pages of data and then abandons the search. I doubt this is very relevant to Postgres. A relational database has to be general purpose, and it's hard to give it "hints"that would tell it when to use this particular optimization. Craig
At 10:52 AM 2/16/2006, Ron wrote: >At 09:48 AM 2/16/2006, Martijn van Oosterhout wrote: > >>Where this does become interesting is where we can convert a datum to >>an integer such that if f(A) > f(B) then A > B. Then we can sort on >>f(X) first with just integer comparisons and then do a full tuple >>comparison only if f(A) = f(B). This would be much more cache-coherent >>and make these algorithms much more applicable in my mind. >In fact we can do better. >Using hash codes or what-not to map datums to keys and then sorting >just the keys and the pointers to those datums followed by an >optional final pass where we do the actual data movement is also a >standard technique for handling large data structures. I thought some follow up might be in order here. Let's pretend that we have the typical DB table where rows are ~2-4KB apiece. 1TB of storage will let us have 256M-512M rows in such a table. A 32b hash code can be assigned to each row value such that only exactly equal rows will have the same hash code. A 32b pointer can locate any of the 256M-512M rows. Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b+32b= 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an optional pass to rearrange the actual rows if we so wish. We get the same result while only examining and manipulating 1/50 to 1/25 as much data during the sort. If we want to spend more CPU time in order to save more space, we can compress the key+pointer representation. That usually reduces the amount of data to be manipulated to ~1/4 the original key+pointer representation, reducing things to ~512M-1GB worth of compressed pointers+keys. Or ~1/200 - ~1/100 the original amount of data we were discussing. Either representation is small enough to fit within RAM rather than requiring HD IO, so we solve the HD IO bottleneck in the best possible way: we avoid ever doing it. Ron
On Thu, Feb 16, 2006 at 11:32:55AM -0500, Ron wrote: > At 10:52 AM 2/16/2006, Ron wrote: > >In fact we can do better. > >Using hash codes or what-not to map datums to keys and then sorting > >just the keys and the pointers to those datums followed by an > >optional final pass where we do the actual data movement is also a > >standard technique for handling large data structures. Or in fact required if the Datums are not all the same size, which is the case in PostgreSQL. > I thought some follow up might be in order here. > > Let's pretend that we have the typical DB table where rows are ~2-4KB > apiece. 1TB of storage will let us have 256M-512M rows in such a table. > > A 32b hash code can be assigned to each row value such that only > exactly equal rows will have the same hash code. > A 32b pointer can locate any of the 256M-512M rows. That hash code is impossible the way you state it, since the set of strings is not mappable to a 32bit integer. You probably meant that a hash code can be assigned such that equal rows have equal hashes (drop the only). > Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b+32b= > 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an optional > pass to rearrange the actual rows if we so wish. > > We get the same result while only examining and manipulating 1/50 to > 1/25 as much data during the sort. But this is what we do now. The tuples are loaded, we sort an array of pointers, then we write the output. Except we don't have the hash, so we require access to the 1TB of data to do the actual comparisons. Even if we did have the hash, we'd *still* need access to the data to handle tie-breaks. That's why your comment about moves always being more expensive than compares makes no sense. A move can be acheived simply by swapping two pointers in the array. A compare actually needs to call all sorts of functions. If and only if we have functions for every data type to produce an ordered hash, we can optimise sorts based on single integers. For reference, look at comparetup_heap(). It's just 20 lines, but each function call there expands to maybe a dozen lines of code. And it has a loop. I don't think we're anywhere near the stage where locality of reference makes much difference. We very rarely needs the tuples actualised in memory in the required order, just the pointers are enough. Have a ncie day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
Attachment
"Craig A. James" <cjames@modgraph-usa.com> writes:
> You can also use this trick when the optimizer is asked for "fastest first result." Say you have a cursor on a
columnof numbers with good distribution. If you do a bucket sort on the first two or three digits only, you know the
first"page" of results will be in the first bucket. So you only need to apply qsort to that first bucket (which is
veryfast, since it's small), and you can deliver the first page of data to the application. This can be particularly
effectivein interactive situations, where the user typically looks at a few pages of data and then abandons the search.
> I doubt this is very relevant to Postgres. A relational database has to be general purpose, and it's hard to give it
"hints"that would tell it when to use this particular optimization.
Actually, LIMIT does nicely for that hint; the PG planner has definitely
got a concept of preferring fast-start plans for limited queries. The
real problem in applying bucket-sort ideas is the lack of any
datatype-independent way of setting up the buckets.
Once or twice we've kicked around the idea of having some
datatype-specific sorting code paths alongside the general-purpose one,
but I can't honestly see this as being workable from a code maintenance
standpoint.
regards, tom lane
On Feb 16, 2006, at 8:32 AM, Ron wrote: > Let's pretend that we have the typical DB table where rows are > ~2-4KB apiece. 1TB of storage will let us have 256M-512M rows in > such a table. > > A 32b hash code can be assigned to each row value such that only > exactly equal rows will have the same hash code. > A 32b pointer can locate any of the 256M-512M rows. > > Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b > +32b= 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an > optional pass to rearrange the actual rows if we so wish. I don't understand this. This is a true statement: (H(x) != H(y)) => (x != y) This is not: (H(x) < H(y)) => (x < y) Hash keys can tell you there's an inequality, but they can't tell you how the values compare. If you want 32-bit keys that compare in the same order as the original values, here's how you have to get them: (1) sort the values into an array (2) use each value's array index as its key It reduces to the problem you're trying to use it to solve. -- Scott Lamb <http://www.slamb.org/>
At 12:19 PM 2/16/2006, Scott Lamb wrote: >On Feb 16, 2006, at 8:32 AM, Ron wrote: >>Let's pretend that we have the typical DB table where rows are >>~2-4KB apiece. 1TB of storage will let us have 256M-512M rows in >>such a table. >> >>A 32b hash code can be assigned to each row value such that only >>exactly equal rows will have the same hash code. >>A 32b pointer can locate any of the 256M-512M rows. >> >>Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b >>+32b= 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an >>optional pass to rearrange the actual rows if we so wish. > >I don't understand this. > >This is a true statement: (H(x) != H(y)) => (x != y) >This is not: (H(x) < H(y)) => (x < y) > >Hash keys can tell you there's an inequality, but they can't tell you >how the values compare. If you want 32-bit keys that compare in the >same order as the original values, here's how you have to get them: For most hash codes, you are correct. There is a class of hash or hash-like codes that maintains the mapping to support that second statement. More later when I can get more time. Ron
On Thu, 2006-02-16 at 12:35 +0100, Steinar H. Gunderson wrote: > glibc-2.3.5/stdlib/qsort.c: > > /* Order size using quicksort. This implementation incorporates > four optimizations discussed in Sedgewick: > > I can't see any references to merge sort in there at all. stdlib/qsort.c defines _quicksort(), not qsort(), which is defined by msort.c. On looking closer, it seems glibc actually tries to determine the physical memory in the machine -- if it is sorting a single array that exceeds 1/4 of the machine's physical memory, it uses quick sort, otherwise it uses merge sort. -Neil
On Thu, 2006-02-16 at 12:15 -0500, Tom Lane wrote: > Once or twice we've kicked around the idea of having some > datatype-specific sorting code paths alongside the general-purpose one, > but I can't honestly see this as being workable from a code maintenance > standpoint. > > regards, tom lane It seems that instead of maintaining a different sorting code path for each data type, you could get away with one generic path and one (hopefully faster) path if you allowed data types to optionally support a 'sortKey' interface by providing a function f which maps inputs to 32- bit int outputs, such that the following two properties hold: f(a)>=f(b) iff a>=b if a==b then f(a)==f(b) So if a data type supports the sortKey interface you could perform the sort on f(value) and only refer back to the actual element comparison functions when two sortKeys have the same value. Data types which could probably provide a useful function for f would be int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). Depending on the overhead, you might not even need to maintain 2 independent search code paths, since you could always use f(x)=0 as the default sortKey function which would degenerate to the exact same sort behavior in use today. -- Mark Lewis
Hi, Mark, Mark Lewis schrieb: > It seems that instead of maintaining a different sorting code path for > each data type, you could get away with one generic path and one > (hopefully faster) path if you allowed data types to optionally support > a 'sortKey' interface by providing a function f which maps inputs to 32- > bit int outputs, such that the following two properties hold: > > f(a)>=f(b) iff a>=b > if a==b then f(a)==f(b) Hmm, to remove redundancy, I'd change the <= to a < and define: if a==b then f(a)==f(b) if a<b then f(a)<=f(b) > Data types which could probably provide a useful function for f would be > int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). With int2 or some restricted ranges of oid and int4, we could even implement a bucket sort. Markus
On Thu, Feb 16, 2006 at 02:17:36PM -0800, Mark Lewis wrote: > It seems that instead of maintaining a different sorting code path for > each data type, you could get away with one generic path and one > (hopefully faster) path if you allowed data types to optionally support > a 'sortKey' interface by providing a function f which maps inputs to 32- > bit int outputs, such that the following two properties hold: > > f(a)>=f(b) iff a>=b > if a==b then f(a)==f(b) Note this is a property of the collation, not the type. For example strings can be sorted in many ways and the sortKey must reflect that. So in postgres terms it's a property of the btree operator class. It's something I'd like to do if I get A Round Tuit. :) Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
Markus Schaber <schabi@logix-tt.com> writes: > Hmm, to remove redundancy, I'd change the <= to a < and define: > > if a==b then f(a)==f(b) > if a<b then f(a)<=f(b) > > > Data types which could probably provide a useful function for f would be > > int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). How exactly do you imagine doing this for text? I could see doing it for char(n)/varchar(n) where n<=4 in SQL_ASCII though. -- greg
> It seems that instead of maintaining a different sorting code path for
> each data type, you could get away with one generic path and one
> (hopefully faster) path if you allowed data types to optionally support
> a 'sortKey' interface by providing a function f which maps inputs to 32-
> bit int outputs, such that the following two properties hold:
Looks like the decorate-sort-undecorate pattern, which works quite well.
Good idea.
I would have said a 64 bit int, but it's the same idea. However it won't
work for floats, which is a pity, because floats fit in 64 bits. Unless
more types creep in the code path (which would not necessarily make it
that slower).
As for text, the worst case is when all strings start with the same 8
letters, but a good case pops up when a few-letter code is used as a key
in a table. Think about a zipcode, for instance. If a merge join needs to
sort on zipcodes, it might as well sort on 64-bits integers...
By the way, I'd like to declare my zipcode columns as SQL_ASCII while the
rest of my database is in UNICODE, so they are faster to index and sort.
Come on, MySQL does it...
Keep up !
On Thu, 2006-02-16 at 17:51 -0500, Greg Stark wrote: > > > Data types which could probably provide a useful function for f would be > > > int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). > > How exactly do you imagine doing this for text? > > I could see doing it for char(n)/varchar(n) where n<=4 in SQL_ASCII though. In SQL_ASCII, just take the first 4 characters (or 8, if using a 64-bit sortKey as elsewhere suggested). The sorting key doesn't need to be a one-to-one mapping. -- Mark Lewis
Hi, PFC, PFC schrieb: > By the way, I'd like to declare my zipcode columns as SQL_ASCII > while the rest of my database is in UNICODE, so they are faster to > index and sort. Come on, MySQL does it... Another use case for parametric column definitions - charset definitions - and the first one that cannot be emulated via constraints. Other use cases I remember were range definitions for numbers or PostGIS dimension, subtype and SRID, but those cann all be emulated via checks / constraints. Markus
On Fri, Feb 17, 2006 at 12:05:23AM +0100, PFC wrote: > I would have said a 64 bit int, but it's the same idea. However it > won't work for floats, which is a pity, because floats fit in 64 bits. Actually, you can compare IEEE floats directly as ints, as long as they're positive. (If they can be both positive and negative, you need to take special care of the sign bit first, but it's still doable.) /* Steinar */ -- Homepage: http://www.sesse.net/
On Thu, 16 Feb 2006, Mark Lewis wrote: > On Thu, 2006-02-16 at 17:51 -0500, Greg Stark wrote: >>>> Data types which could probably provide a useful function for f would be >>>> int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). >> >> How exactly do you imagine doing this for text? >> >> I could see doing it for char(n)/varchar(n) where n<=4 in SQL_ASCII though. > > > In SQL_ASCII, just take the first 4 characters (or 8, if using a 64-bit > sortKey as elsewhere suggested). The sorting key doesn't need to be a > one-to-one mapping. that would violate your second contraint ( f(a)==f(b) iff (a==b) ) if you could drop that constraint (the cost of which would be extra 'real' compares within a bucket) then a helper function per datatype could work as you are talking. David Lang
At 01:47 PM 2/16/2006, Ron wrote: >At 12:19 PM 2/16/2006, Scott Lamb wrote: >>On Feb 16, 2006, at 8:32 AM, Ron wrote: >>>Let's pretend that we have the typical DB table where rows are >>>~2-4KB apiece. 1TB of storage will let us have 256M-512M rows in >>>such a table. >>> >>>A 32b hash code can be assigned to each row value such that only >>>exactly equal rows will have the same hash code. >>>A 32b pointer can locate any of the 256M-512M rows. >>> >>>Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b >>>+32b= 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an >>>optional pass to rearrange the actual rows if we so wish. >> >>I don't understand this. >> >>This is a true statement: (H(x) != H(y)) => (x != y) >>This is not: (H(x) < H(y)) => (x < y) >> >>Hash keys can tell you there's an inequality, but they can't tell you >>how the values compare. If you want 32-bit keys that compare in the >>same order as the original values, here's how you have to get them: >For most hash codes, you are correct. There is a class of hash or >hash-like codes that maintains the mapping to support that second statement. > >More later when I can get more time. >Ron OK, so here's _a_ way (there are others) to obtain a mapping such that if a < b then f(a) < f (b) and if a == b then f(a) == f(b) Pretend each row is a integer of row size (so a 2KB row becomes a 16Kb integer; a 4KB row becomes a 32Kb integer; etc) Since even a 1TB table made of such rows can only have 256M - 512M possible values even if each row is unique, a 28b or 29b key is large enough to represent each row's value and relative rank compared to all of the others even if all row values are unique. By scanning the table once, we can map say 0000001h (Hex used to ease typing) to the row with the minimum value and 1111111h to the row with the maximum value as well as mapping everything in between to their appropriate keys. That same scan can be used to assign a pointer to each record's location. We can now sort the key+pointer pairs instead of the actual data and use an optional final pass to rearrange the actual rows if we wish. That initial scan to set up the keys is expensive, but if we wish that cost can be amortized over the life of the table so we don't have to pay it all at once. In addition, once we have created those keys, then can be saved for later searches and sorts. Further space savings can be obtained whenever there are duplicate keys and/or when compression methods are used on the Key+pointer pairs. Ron
On fös, 2006-02-17 at 01:20 -0500, Ron wrote: > At 01:47 PM 2/16/2006, Ron wrote: > >At 12:19 PM 2/16/2006, Scott Lamb wrote: > >>On Feb 16, 2006, at 8:32 AM, Ron wrote: > >>>Let's pretend that we have the typical DB table where rows are > >>>~2-4KB apiece. 1TB of storage will let us have 256M-512M rows in > >>>such a table. > >>> > >>>A 32b hash code can be assigned to each row value such that only > >>>exactly equal rows will have the same hash code. > >>>A 32b pointer can locate any of the 256M-512M rows. > >>> > >>>Now instead of sorting 1TB of data we can sort 2^28 to 2^29 32b > >>>+32b= 64b*(2^28 to 2^29)= 2-4GB of pointers+keys followed by an > >>>optional pass to rearrange the actual rows if we so wish. > >> > >>I don't understand this. > >> > >>This is a true statement: (H(x) != H(y)) => (x != y) > >>This is not: (H(x) < H(y)) => (x < y) > >> > >>Hash keys can tell you there's an inequality, but they can't tell you > >>how the values compare. If you want 32-bit keys that compare in the > >>same order as the original values, here's how you have to get them: > >For most hash codes, you are correct. There is a class of hash or > >hash-like codes that maintains the mapping to support that second statement. > > > >More later when I can get more time. > >Ron > > OK, so here's _a_ way (there are others) to obtain a mapping such that > if a < b then f(a) < f (b) and > if a == b then f(a) == f(b) > By scanning the table once, we can map say 0000001h (Hex used to ease > typing) to the row with the minimum value and 1111111h to the row > with the maximum value as well as mapping everything in between to > their appropriate keys. That same scan can be used to assign a > pointer to each record's location. This step is just as expensive as the original sort you want to replace/improve. If you want to keep this mapping saved as a sort of an index, or as part ot each row data, this will make the cost of inserts and updates enormous. > > We can now sort the key+pointer pairs instead of the actual data and > use an optional final pass to rearrange the actual rows if we wish. How are you suggesting this mapping be accessed? If the mapping is kept separate from the tuple data, as in an index, then how will you look up the key? > That initial scan to set up the keys is expensive, but if we wish > that cost can be amortized over the life of the table so we don't > have to pay it all at once. In addition, once we have created those > keys, then can be saved for later searches and sorts. What is the use case where this would work better than a regular btree index ? gnari
Hi, David, David Lang schrieb: >> In SQL_ASCII, just take the first 4 characters (or 8, if using a 64-bit >> sortKey as elsewhere suggested). The sorting key doesn't need to be a >> one-to-one mapping. > that would violate your second contraint ( f(a)==f(b) iff (a==b) ) no, it doesn't. When both strings are equal, then the first characters are equal, too. If they are not equal, the constraint condition does not match. The first characters of the strings may be equal as f(a) may be equal to f(b) as to the other constraint. Markus
Hi, Ron, Ron schrieb: > OK, so here's _a_ way (there are others) to obtain a mapping such that > if a < b then f(a) < f (b) and > if a == b then f(a) == f(b) > > Pretend each row is a integer of row size (so a 2KB row becomes a 16Kb > integer; a 4KB row becomes a 32Kb integer; etc) > Since even a 1TB table made of such rows can only have 256M - 512M > possible values even if each row is unique, a 28b or 29b key is large > enough to represent each row's value and relative rank compared to all > of the others even if all row values are unique. > > By scanning the table once, we can map say 0000001h (Hex used to ease > typing) to the row with the minimum value and 1111111h to the row with > the maximum value as well as mapping everything in between to their > appropriate keys. That same scan can be used to assign a pointer to > each record's location. But with a single linear scan, this cannot be accomplished, as the table contents are neither sorted nor distributed linearly between the minimum and the maximum. For this mapping, you need a full table sort. > That initial scan to set up the keys is expensive, but if we wish that > cost can be amortized over the life of the table so we don't have to pay > it all at once. In addition, once we have created those keys, then can > be saved for later searches and sorts. But for every update or insert, you have to resort the keys, which is _very_ expensive as it basically needs to update a huge part of the table. Markus
Has anybody got some profiler data on the amount of time spent in
comparisons during a sort ? Say, the proposals here would give the most
gains on simple types like INTEGER ; so it would be interesting to know
how much time is now spent in comparisons for sorting a column of ints. If
it's like, 10% of the total time, well...
More hand-waving :
What are the usage case for sorts ?
- potentially huge data sets : create index, big joins, reporting queries
etc.
- small data sets : typically, a query with an ORDER BY which will return
a small amount of rows (website stuff), or joins not small enough to use a
HashAggregate, but not big enough to create an index just for them.
The cost of a comparison vs. moving stuff around and fetching stuff is
probably very different in these two cases. If it all neatly fits in
sort_mem, you can do fancy stuff (like sorting on SortKeys) which will
need to access the data in almost random order when time comes to hand the
sorted data back. So, I guess the SortKey idea would rather apply to the
latter case only, which is CPU limited.
Anyway, I was wondering about queries with multipart keys, like ORDER BY
zipcode, client_name, date and the like. Using just an int64 as the key
isn't going to help a lot here. Why not use a binary string of limited
length ? I'd tend to think it would not be that slower than comparing
ints, and it would be faster than calling each comparison function for
each column. Each key part would get assigned to a byte range in the
string.
It would waste some memory, but for instance, using 2x the memory for
half the time would be a good tradeoff if the amount of memory involved is
in the order of megabytes.
Also, it would solve the problem of locales. Comparisons involving
locales are slow, but strings can be converted to a canonical form
(removing accents and stuff), and then binary sorted.
Also I'll insert a plug for the idea that the Sort needs to know if there
will be a LIMIT afterwards ; this way it could reduce its working set by
simply discarding the rows which would have been discarded anyway by the
LIMIT. Say you want the 100 first rows out of a million ordered rows. If
the sort knows this, it can be performed in the amount of memory for a 100
rows sort.
At 04:24 AM 2/17/2006, Ragnar wrote: >On fös, 2006-02-17 at 01:20 -0500, Ron wrote: > > > > OK, so here's _a_ way (there are others) to obtain a mapping such that > > if a < b then f(a) < f (b) and > > if a == b then f(a) == f(b) > > > By scanning the table once, we can map say 0000001h (Hex used to ease > > typing) to the row with the minimum value and 1111111h to the row > > with the maximum value as well as mapping everything in between to > > their appropriate keys. That same scan can be used to assign a > > pointer to each record's location. > >This step is just as expensive as the original >sort you want to replace/improve. Why do you think that? External sorts involve the equivalent of multiple scans of the table to be sorted, sometimes more than lgN (where N is the number of items in the table to be sorted). Since this is physical IO we are talking about, each scan is very expensive, and therefore 1 scan is going to take considerably less time than >= lgN scans will be. >If you want to keep this mapping saved as a sort >of an index, or as part ot each row data, this >will make the cost of inserts and updates enormous. Not sure you've got this right either. Looks to me like we are adding a <= 32b quantity to each row. Once we know the mapping, incrementally updating it upon insert or update would seem to be simple matter of a fast search for the correct ranking [Interpolation search, which we have all the needed data for, is O(lglgN). Hash based search is O(1)]; plus an increment/decrement of the key values greater/less than the key value of the row being inserted / updated. Given than we are updating all the keys in a specific range within a tree structure, that update can be done in O(lgm) (where m is the number of records affected). > > We can now sort the key+pointer pairs instead of the actual data and > > use an optional final pass to rearrange the actual rows if we wish. > >How are you suggesting this mapping be accessed? >If the mapping is kept separate from the tuple >data, as in an index, then how will you look up the key? ??? We've effectively created a data set where each record is a pointer to a DB row plus its key. We can now sort the data set by key and then do an optional final pass to rearrange the actual DB rows if we so wish. Since that final pass is very expensive, it is good that not all use scenarios will need that final pass. The amount of storage required to sort this representation of the table rather than the actual table is so much less that it turns an external sorting problem into a internal sorting problem with an optional final pass that is =1= scan (albeit one scan with a lot of seeks and data movement). This is a big win. It is a variation of a well known technique. See Sedgewick, Knuth, etc. > > That initial scan to set up the keys is expensive, but if we wish > > that cost can be amortized over the life of the table so we don't > > have to pay it all at once. In addition, once we have created those > > keys, then can be saved for later searches and sorts. > >What is the use case where this would work better than a >regular btree index ? Again, ??? btree indexes address different issues. They do not in any way help create a compact data representation of the original data that saves enough space so as to turn an external ranking or sorting problem into an internal one. Ron
At 05:19 AM 2/17/2006, Markus Schaber wrote: >Hi, Ron, > >Ron schrieb: > > > OK, so here's _a_ way (there are others) to obtain a mapping such that > > if a < b then f(a) < f (b) and > > if a == b then f(a) == f(b) > > > > Pretend each row is a integer of row size (so a 2KB row becomes a 16Kb > > integer; a 4KB row becomes a 32Kb integer; etc) > > Since even a 1TB table made of such rows can only have 256M - 512M > > possible values even if each row is unique, a 28b or 29b key is large > > enough to represent each row's value and relative rank compared to all > > of the others even if all row values are unique. > > > > By scanning the table once, we can map say 0000001h (Hex used to ease > > typing) to the row with the minimum value and 1111111h to the row with > > the maximum value as well as mapping everything in between to their > > appropriate keys. That same scan can be used to assign a pointer to > > each record's location. > >But with a single linear scan, this cannot be accomplished, as the table >contents are neither sorted nor distributed linearly between the minimum >and the maximum. So what? We are talking about key assignment here, not anything that requires physically manipulating the actual DB rows. One physical IO pass should be all that's needed. >For this mapping, you need a full table sort. One physical IO pass should be all that's needed. However, let's pretend you are correct and that we do need to sort the table to get the key mapping. Even so, we would only need to do it =once= and then we would be able to use and incrementally update the results forever afterward. Even under this assumption, one external sort to save all subsequent such sorts seems well worth it. IOW, even if I'm wrong about the initial cost to do this; it is still worth doing ;-) > > That initial scan to set up the keys is expensive, but if we wish that > > cost can be amortized over the life of the table so we don't have to pay > > it all at once. In addition, once we have created those keys, then can > > be saved for later searches and sorts. > >But for every update or insert, you have to resort the keys, which is >_very_ expensive as it basically needs to update a huge part of the table. ??? You do not need to resort already ordered data to insert a new element into it such that the data stays ordered! Once we have done the key ordering operation once, we should not ever need to do it again on the original data. Else most sorting algorithms wouldn't work ;-) Ron
On Fri, Feb 17, 2006 at 08:23:40AM -0500, Ron wrote: > >For this mapping, you need a full table sort. > One physical IO pass should be all that's needed. However, let's > pretend you are correct and that we do need to sort the table to get > the key mapping. Even so, we would only need to do it =once= and > then we would be able to use and incrementally update the results > forever afterward. Even under this assumption, one external sort to > save all subsequent such sorts seems well worth it. > > IOW, even if I'm wrong about the initial cost to do this; it is still > worth doing ;-) I think you're talking about something different here. You're thinking of having the whole table sorted and when you add a new value you add it in such a way to keep it sorted. The problem is, what do you sort it by? If you've sorted the table by col1, then when the user does ORDER BY col2 it's useless. Indeed, this is what btrees do, you store the order of the table seperate from the data. And you can store multiple orders. But even then, when someone does ORDER BY lower(col1), it's still useless. And you're right, we still need to do the single mass sort in the beginning, which is precisely what we're trying to optimise here. > ??? You do not need to resort already ordered data to insert a new > element into it such that the data stays ordered! Once we have done > the key ordering operation once, we should not ever need to do it > again on the original data. Else most sorting algorithms wouldn't work ;-) We already do this with btree indexes. I'm not sure what you are proposing that improves on that. Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
On Feb 16, 2006, at 2:17 PM, Mark Lewis wrote:
Data types which could probably provide a useful function for f would be
int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII).
...and with some work, floats (I think just the exponent would work, if nothing else). bytea. Probably just about anything.
Interesting. If you abandon the idea that collisions should be impossible (they're not indexes) or extremely rare (they're not hashes), it's pretty easy to come up with a decent hint to avoid a lot of dereferences.
--
Scott Lamb <http://www.slamb.org/>
On Fri, Feb 17, 2006 at 08:18:41AM -0800, Scott Lamb wrote: > On Feb 16, 2006, at 2:17 PM, Mark Lewis wrote: > >Data types which could probably provide a useful function for f > >would be > >int2, int4, oid, and possibly int8 and text (at least for SQL_ASCII). > > ...and with some work, floats (I think just the exponent would work, > if nothing else). bytea. Probably just about anything. > > Interesting. If you abandon the idea that collisions should be > impossible (they're not indexes) or extremely rare (they're not > hashes), it's pretty easy to come up with a decent hint to avoid a > lot of dereferences. Yep, pretty much for any datatype you create a mapping function to map it to a signed int32. All you have to guarentee is that f(a) > f(b) implies that a > b. Only if f(a) == f(b) do you need to compare a and b. You then change the sorting code to have an array of (Datum,int32) (ouch, double the storage) where the int32 is the f(Datum). And in the comparison routines you first check the int32. If they give an order you're done. On match you do the full comparison. For integer types (int2,int4,int8,oid) the conversion is straight forward. For float you'd use the exponent and the first few bits of the mantissa. For strings you'd have to bail, or use a strxfrm equivalent. NULL would be INT_MAX pr INT_MIN depending on where you want it. Thing is, even if you don't have such a function and always return zero, the results will still be right. Not a new idea, but it would be very nice to implement. If would produce nice speedups for type where comparisons are expensive. And more importantly, the bulk of the comparisons can be moved inline and make the whole cache-friendlyness discussed here much more meaningful. Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Patent. n. Genius is 5% inspiration and 95% perspiration. A patent is a > tool for doing 5% of the work and then sitting around waiting for someone > else to do the other 95% so you can sue them.
Attachment
At 10:53 AM 2/17/2006, Martijn van Oosterhout wrote: >On Fri, Feb 17, 2006 at 08:23:40AM -0500, Ron wrote: > > >For this mapping, you need a full table sort. > > One physical IO pass should be all that's needed. However, let's > > pretend you are correct and that we do need to sort the table to get > > the key mapping. Even so, we would only need to do it =once= and > > then we would be able to use and incrementally update the results > > forever afterward. Even under this assumption, one external sort to > > save all subsequent such sorts seems well worth it. > > > > IOW, even if I'm wrong about the initial cost to do this; it is still > > worth doing ;-) > >I think you're talking about something different here. You're thinking >of having the whole table sorted and when you add a new value you add >it in such a way to keep it sorted. The problem is, what do you sort it >by? If you've sorted the table by col1, then when the user does ORDER >BY col2 it's useless. No, I'm thinking about how to represent DB row data in such a way that a= we use a compact enough representation that we can sort internally rather than externally. b= we do the sort once and avoid most of the overhead upon subsequent similar requests. I used the example of sorting on the entire row to show that the approach works even when the original record being sorted by is very large. All my previous comments on this topic hold for the case where we are sorting on only part of a row as well. If all you are doing is sorting on a column or a few columns, what I'm discussing is even easier since treating the columns actually being used a sort criteria as integers rather than the whole row as an atomic unit eats less resources during the key creation and mapping process. If the row is 2-4KB in size, but we only care about some combination of columns that only takes on <= 2^8 or <= 2^16 different values, then what I've described will be even better than the original example I gave. Basically, the range of a key is going to be restricted by how a= big the field is that represents the key (columns and such are usually kept narrow for performance reasons) or b= big each row is (the more space each row takes, the fewer rows fit into any given amount of storage) c= many rows there are in the table Between the conditions, the range of a key tends to be severely restricted and therefore use much less space than sorting the actual DB records would. ...and that gives us something we can take advantage of. >Indeed, this is what btrees do, you store the order of the table >seperate from the data. And you can store multiple orders. But even >then, when someone does ORDER BY lower(col1), it's still useless. > >And you're right, we still need to do the single mass sort in the >beginning, which is precisely what we're trying to optimise here. Sigh. My points were: 1= we have information available to us that allows us to map the rows in such a way as to turn most external sorts into internal sorts, thereby avoiding the entire external sorting problem in those cases. This is a huge performance improvement. 2= that an external sort is =NOT= required for initial key assignment, but even if it was it would be worth it. 3= that this is a variation of a well known technique so I'm not suggesting heresy here. Ron
On Thu, 2006-02-16 at 21:33 -0800, David Lang wrote: > > In SQL_ASCII, just take the first 4 characters (or 8, if using a 64-bit > > sortKey as elsewhere suggested). The sorting key doesn't need to be a > > one-to-one mapping. > > that would violate your second contraint ( f(a)==f(b) iff (a==b) ) > > if you could drop that constraint (the cost of which would be extra 'real' > compares within a bucket) then a helper function per datatype could work > as you are talking. I think we're actually on the same page here; you're right that the constraint above ( f(a)==f(b) iff a==b ) can't be extended to data types with more than 32 bits of value space. But the constraint I listed was actually: if a==b then f(a)==f(b) Which doesn't imply 'if and only if'. It's a similar constraint to hashcodes; the same value will always have the same hash, but you're not guaranteed that the hashcodes for two distinct values will be unique. -- Mark
On fös, 2006-02-17 at 08:01 -0500, Ron wrote: > At 04:24 AM 2/17/2006, Ragnar wrote: > >On fös, 2006-02-17 at 01:20 -0500, Ron wrote: > > > > > > OK, so here's _a_ way (there are others) to obtain a mapping such that > > > if a < b then f(a) < f (b) and > > > if a == b then f(a) == f(b) > > > > > By scanning the table once, we can map say 0000001h (Hex used to ease > > > typing) to the row with the minimum value and 1111111h to the row > > > with the maximum value as well as mapping everything in between to > > > their appropriate keys. That same scan can be used to assign a > > > pointer to each record's location. > > > >This step is just as expensive as the original > >sort you want to replace/improve. > > Why do you think that? External sorts involve > the equivalent of multiple scans of the table to > be sorted, sometimes more than lgN (where N is > the number of items in the table to be > sorted). Since this is physical IO we are > talking about, each scan is very expensive, and > therefore 1 scan is going to take considerably > less time than >= lgN scans will be. Call me dim, but please explain exactly how you are going to build this mapping in one scan. Are you assuming the map will fit in memory? > > > >If you want to keep this mapping saved as a sort > >of an index, or as part ot each row data, this > >will make the cost of inserts and updates enormous. > > Not sure you've got this right either. Looks to > me like we are adding a <= 32b quantity to each > row. Once we know the mapping, incrementally > updating it upon insert or update would seem to > be simple matter of a fast search for the correct > ranking [Interpolation search, which we have all > the needed data for, is O(lglgN). Hash based > search is O(1)]; plus an increment/decrement of > the key values greater/less than the key value of > the row being inserted / updated. Given than we > are updating all the keys in a specific range > within a tree structure, that update can be done > in O(lgm) (where m is the number of records affected). Say again ? Let us say you have 1 billion rows, where the column in question contains strings like baaaaaaaaaaaaaaa....aaa baaaaaaaaaaaaaaa....aab baaaaaaaaaaaaaaa....aac ... not necessarily in this order on disc of course The minimum value would be keyed as 00000001h, the next one as 00000002h and so on. Now insert new value 'aaaaa' Not only will you have to update 1 billion records, but also all the values in your map. please explain gnari
Mark Lewis <mark.lewis@mir3.com> writes:
> I think we're actually on the same page here; you're right that the
> constraint above ( f(a)==f(b) iff a==b ) can't be extended to data types
> with more than 32 bits of value space. But the constraint I listed was
> actually:
> if a==b then f(a)==f(b)
I believe Martijn had it right: the important constraint is
f(a) > f(b) implies a > b
which implies by commutativity
f(a) < f(b) implies a < b
and these two together imply
a == b implies f(a) == f(b)
Now you can't do any sorting if you only have the equality rule, you
need the inequality rule.
regards, tom lane
Last night I implemented a non-recursive introsort in C... let me test it a bit more and then I'll post it here for everyone else to try out.
--
Jonah H. Harris, Database Internals Architect
EnterpriseDB Corporation
732.331.1324
On 2/16/06, Markus Schaber <schabi@logix-tt.com> wrote:
Hi, Ron,
Ron wrote:
> ...and of course if you know enough about the data to be sorted so as to
> constrain it appropriately, one should use a non comparison based O(N)
> sorting algorithm rather than any of the general comparison based
> O(NlgN) methods.
Sounds interesting, could you give us some pointers (names, URLs,
papers) to such algorithms?
Thanks a lot,
Markus
--
Markus Schaber | Logical Tracking&Tracing International AG
Dipl. Inf. | Software Development GIS
Fight against software patents in EU! www.ffii.org www.nosoftwarepatents.org
---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?
http://archives.postgresql.org
--
Jonah H. Harris, Database Internals Architect
EnterpriseDB Corporation
732.331.1324
On 2/17/06, Ragnar <gnari@hive.is> wrote: > Say again ? > Let us say you have 1 billion rows, where the > column in question contains strings like > baaaaaaaaaaaaaaa....aaa > baaaaaaaaaaaaaaa....aab > baaaaaaaaaaaaaaa....aac > ... > not necessarily in this order on disc of course > > The minimum value would be keyed as 00000001h, > the next one as 00000002h and so on. > > Now insert new value 'aaaaa' > > Not only will you have to update 1 billion records, > but also all the values in your map. > > please explain No comment on the usefulness of the idea overall.. but the solution would be to insert with the colliding value of the existing one lesser than it.. It will falsly claim equal, which you then must fix with a second local sort which should be fast because you only need to sort the duplicates/false dupes. If you insert too much then this obviously becomes completely useless.
Last month I wrote: > It seems clear that our qsort.c is doing a pretty awful job of picking > qsort pivots, while glibc is mostly managing not to make that mistake. I re-ran Gary's test script using the just-committed improvements to qsort.c, and got pretty nice numbers (attached --- compare to http://archives.postgresql.org/pgsql-performance/2006-02/msg00227.php). So it was wrong to blame his problems on the pivot selection --- the culprit was that ill-considered switch to insertion sort. regards, tom lane 100 runtimes for latest port/qsort.c, sorted ascending: Time: 335.481 ms Time: 335.606 ms Time: 335.932 ms Time: 336.039 ms Time: 336.182 ms Time: 336.231 ms Time: 336.711 ms Time: 336.721 ms Time: 336.971 ms Time: 336.982 ms Time: 337.036 ms Time: 337.190 ms Time: 337.223 ms Time: 337.312 ms Time: 337.350 ms Time: 337.423 ms Time: 337.523 ms Time: 337.528 ms Time: 337.565 ms Time: 337.566 ms Time: 337.732 ms Time: 337.741 ms Time: 337.744 ms Time: 337.786 ms Time: 337.790 ms Time: 337.898 ms Time: 337.905 ms Time: 337.952 ms Time: 337.976 ms Time: 338.017 ms Time: 338.123 ms Time: 338.206 ms Time: 338.306 ms Time: 338.514 ms Time: 338.594 ms Time: 338.597 ms Time: 338.683 ms Time: 338.705 ms Time: 338.729 ms Time: 338.748 ms Time: 338.816 ms Time: 338.958 ms Time: 338.963 ms Time: 338.997 ms Time: 339.074 ms Time: 339.106 ms Time: 339.134 ms Time: 339.159 ms Time: 339.226 ms Time: 339.260 ms Time: 339.289 ms Time: 339.341 ms Time: 339.500 ms Time: 339.585 ms Time: 339.595 ms Time: 339.774 ms Time: 339.897 ms Time: 339.927 ms Time: 340.064 ms Time: 340.133 ms Time: 340.172 ms Time: 340.219 ms Time: 340.261 ms Time: 340.323 ms Time: 340.708 ms Time: 340.761 ms Time: 340.785 ms Time: 340.900 ms Time: 340.986 ms Time: 341.339 ms Time: 341.564 ms Time: 341.707 ms Time: 342.155 ms Time: 342.213 ms Time: 342.452 ms Time: 342.515 ms Time: 342.540 ms Time: 342.928 ms Time: 343.548 ms Time: 343.663 ms Time: 344.192 ms Time: 344.952 ms Time: 345.152 ms Time: 345.174 ms Time: 345.444 ms Time: 346.848 ms Time: 348.144 ms Time: 348.842 ms Time: 354.550 ms Time: 356.877 ms Time: 357.475 ms Time: 358.487 ms Time: 364.178 ms Time: 370.730 ms Time: 493.098 ms Time: 648.009 ms Time: 849.345 ms Time: 860.616 ms Time: 936.800 ms Time: 1727.085 ms
I'm reposting this -- I sent this out a month ago but never got a response, and hope someone can shed some light on
this.
Thanks,
Craig
--------------------------
This is a straightforward query that should be fairly quick, but takes about 30 minutes. It's a query across three
tables,call them A, B, and C. The tables are joined on indexed columns.
Here's a quick summary:
Table A -----> Table B -----> Table C
A_ID B_ID C_ID
A_ID NAME
C_ID
Tables A and B have 6 million rows each. Table C is small: 67 names, no repeats. All columns involved in the join are
indexed. The database has been full-vacuumed and analyzed.
Summary:
1. Query B only: 2.7 seconds, 302175 rows returned
2. Join B and C: 4.3 seconds, exact same answer
3. Join A and B: 7.2 minutes, exact same answer
4. Join A, B, C: 32.7 minutes, exact same answer
Looking at these:
Query #1 is doing the real work: finding the rows of interest.
Queries #1 and #2 ought to be virtually identical, since Table C has
just one row with C_ID = 9, but the time almost doubles.
Query #3 should take a bit longer than Query #1 because it has to join
300K rows, but the indexes should make this take just a few seconds,
certainly well under a minute.
Query #4 should be identical to Query #3, again because there's only
one row in Table C. 32 minutes is pretty horrible for such a
straightforward query.
It looks to me like the problem is the use of nested loops when a hash join should be used, but I'm no expert at query
planning.
This is psql 8.0.3. Table definitions are at the end. Hardware is a Dell, 2-CPU Xeon, 4 GB memory, database is on a
singleSATA 7200RPM disk.
These table and column names are altered to protect the guilty, otherwise these are straight from Postgres.
QUERY #1:
---------
explain analyze select B.A_ID from B where B.B_ID = 9;
Index Scan using i_B_B_ID on B (cost=0.00..154401.36 rows=131236 width=4) (actual time=0.158..1387.251 rows=302175
loops=1)
Index Cond: (B_ID = 9)
Total runtime: 2344.053 ms
QUERY #2:
---------
explain analyze select B.A_ID from B join C on (B.C_ID = C.C_ID) where C.name = 'Joe';
Nested Loop (cost=0.00..258501.92 rows=177741 width=4) (actual time=0.349..3392.532 rows=302175 loops=1)
-> Seq Scan on C (cost=0.00..12.90 rows=1 width=4) (actual time=0.232..0.336 rows=1 loops=1)
Filter: ((name)::text = 'Joe'::text)
-> Index Scan using i_B_C_ID on B (cost=0.00..254387.31 rows=328137 width=8) (actual time=0.102..1290.002 rows=302175
loops=1)
Index Cond: (B.C_ID = "outer".C_ID)
Total runtime: 4373.916 ms
QUERY #3:
---------
explain analyze
select A.A_ID from A
join B on (A.A_ID = B.A_ID) where B.B_ID = 9;
Nested Loop (cost=0.00..711336.41 rows=131236 width=4) (actual time=37.118..429419.347 rows=302175 loops=1)
-> Index Scan using i_B_B_ID on B (cost=0.00..154401.36 rows=131236 width=4) (actual time=27.344..8858.489
rows=302175loops=1)
Index Cond: (B_ID = 9)
-> Index Scan using pk_A_test on A (cost=0.00..4.23 rows=1 width=4) (actual time=1.372..1.376 rows=1 loops=302175)
Index Cond: (A.A_ID = "outer".A_ID)
Total runtime: 430467.686 ms
QUERY #4:
---------
explain analyze
select A.A_ID from A
join B on (A.A_ID = B.A_ID)
join C on (B.B_ID = C.B_ID)
where C.name = 'Joe';
Nested Loop (cost=0.00..1012793.38 rows=177741 width=4) (actual time=70.184..1960112.247 rows=302175 loops=1)
-> Nested Loop (cost=0.00..258501.92 rows=177741 width=4) (actual time=52.114..17753.638 rows=302175 loops=1)
-> Seq Scan on C (cost=0.00..12.90 rows=1 width=4) (actual time=0.109..0.176 rows=1 loops=1)
Filter: ((name)::text = 'Joe'::text)
-> Index Scan using i_B_B_ID on B (cost=0.00..254387.31 rows=328137 width=8) (actual time=51.985..15566.896
rows=302175loops=1)
Index Cond: (B.B_ID = "outer".B_ID)
-> Index Scan using pk_A_test on A (cost=0.00..4.23 rows=1 width=4) (actual time=6.407..6.412 rows=1 loops=302175)
Index Cond: (A.A_ID = "outer".A_ID)
Total runtime: 1961200.079 ms
TABLE DEFINITIONS:
------------------
xxx => \d a
Table "xxx.a"
Column | Type | Modifiers
------------------+------------------------+-----------
a_id | integer | not null
... more columns
Indexes:
"pk_a_id" PRIMARY KEY, btree (a_id)
... more indexes on other columns
xxx => \d b
Table "xxx.b"
Column | Type | Modifiers
-------------------------+------------------------+-----------
b_id | integer | not null
a_id | integer | not null
c_id | integer | not null
... more columns
Indexes:
"b_pkey" PRIMARY KEY, btree (b_id)
"i_b_a_id" btree (a_id)
"i_b_c_id" btree (c_id)
xxx=> \d c
Table "xxx.c"
Column | Type | Modifiers
--------------+------------------------+-----------
c_id | integer | not null
name | character varying(200) |
... more columns
Indexes:
"c_pkey" PRIMARY KEY, btree (c_id)
"Craig A. James" <cjames@modgraph-usa.com> writes:
> It looks to me like the problem is the use of nested loops when a hash join should be used, but I'm no expert at
queryplanning.
Given the sizes of the tables involved, you'd likely have to boost up
work_mem before the planner would consider a hash join. What nondefault
configuration settings do you have, anyway?
regards, tom lane
Tom Lane wrote: > "Craig A. James" <cjames@modgraph-usa.com> writes: >> It looks to me like the problem is the use of nested loops when a hash >> join should be used, but I'm no expert at query planning. > > Given the sizes of the tables involved, you'd likely have to boost up > work_mem before the planner would consider a hash join. What nondefault > configuration settings do you have, anyway? shared_buffers = 20000 work_mem = 32768 effective_cache_size = 300000 This is on a 4GB machine. Is there a guideline for work_mem that's related to table size? Something like, "allow 2 MB permillion rows"? I'm also curious why the big difference between my "Query #1" and "Query #2". Even though it does a nested loop, #2's outerloop only returns one result from a very tiny table, so shouldn't it be virtually indistinguishable from #1? Thanks, Craig
"Craig A. James" <cjames@modgraph-usa.com> writes:
> Tom Lane wrote:
>> Given the sizes of the tables involved, you'd likely have to boost up
>> work_mem before the planner would consider a hash join. What nondefault
>> configuration settings do you have, anyway?
> shared_buffers = 20000
> work_mem = 32768
> effective_cache_size = 300000
So for a 6M-row table, 32M work_mem would allow ... um ... 5 bytes per
row. It's not happening :-(
Try boosting work_mem by a factor of 100 and seeing whether a hash-based
join actually wins or not. If so, we can discuss where the sane setting
really falls, if not there's no point.
regards, tom lane
On Tue, Mar 21, 2006 at 05:04:16PM -0800, Craig A. James wrote: > Tom Lane wrote: > >"Craig A. James" <cjames@modgraph-usa.com> writes: > >>It looks to me like the problem is the use of nested loops when a hash > >>join should be used, but I'm no expert at query planning. > > > >Given the sizes of the tables involved, you'd likely have to boost up > >work_mem before the planner would consider a hash join. What nondefault > >configuration settings do you have, anyway? > > shared_buffers = 20000 > work_mem = 32768 > effective_cache_size = 300000 > > This is on a 4GB machine. Is there a guideline for work_mem that's related > to table size? Something like, "allow 2 MB per million rows"? No. The general guide is "set it as large as possible without making the machine start swapping." In some cases, you'll want to bump it up much higher for certain queries, especially if you know those queries will only run one at a time. -- Jim C. Nasby, Sr. Engineering Consultant jnasby@pervasive.com Pervasive Software http://pervasive.com work: 512-231-6117 vcard: http://jim.nasby.net/pervasive.vcf cell: 512-569-9461