Thread: Joshua Drake's "What MySQL Can Learn From PostgreSQL" talk at MySQL Con 2008

JD's talk and slides:
http://www.commandprompt.com/blogs/joshua_drake/2008/04/what_mysql_and_really_sun_can_learn_from_postgresql/

Review of the talk:
http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql


--
Selena Deckelmann
United States PostgreSQL Association - http://www.postgresql.us
PDXPUG - http://pugs.postgresql.org/pdx
Me - http://www.chesnok.com/daily

On Fri, 2008-04-18 at 10:38 -0700, Selena Deckelmann wrote:
> JD's talk and slides:
> http://www.commandprompt.com/blogs/joshua_drake/2008/04/what_mysql_and_really_sun_can_learn_from_postgresql/
>
> Review of the talk:
> http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql
>

Interesting reads.

I have to agree with the part about lack of in place upgrades.  I would
have say that that is the single biggest thorn in our PG using side here
at Afilias.  Upgrading the data layer via Slony is a **massive**
undertaking.

Brad.

On Fri, 18 Apr 2008 14:03:53 -0400
Brad Nicholson <bnichols@ca.afilias.info> wrote:

> On Fri, 2008-04-18 at 10:38 -0700, Selena Deckelmann wrote:
> > JD's talk and slides:
> > http://www.commandprompt.com/blogs/joshua_drake/2008/04/what_mysql_and_really_sun_can_learn_from_postgresql/
> >
> > Review of the talk:
> > http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql
> >
>
> Interesting reads.
>
> I have to agree with the part about lack of in place upgrades.  I
> would have say that that is the single biggest thorn in our PG using
> side here at Afilias.  Upgrading the data layer via Slony is a
> **massive** undertaking.

The problem is not that it is inappropriate. It is that it doesn't work
for databases of that size. Even at 500GB you can have serious issues
with Slony.

This is not actually a Slony limitation but a PostgreSQL one.

Sincerely,

Joshua D. Drake


--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate

On Fri, Apr 18, 2008 at 02:03:53PM -0400, Brad Nicholson wrote:
> at Afilias.  Upgrading the data layer via Slony is a **massive**
> undertaking.

While this is true, I can tell you from unpleasant experience that it's way
less hard than it was without Slony ;-)

A

On Fri, 2008-04-18 at 11:13 -0700, Joshua D. Drake wrote:
> On Fri, 18 Apr 2008 14:03:53 -0400
> Brad Nicholson <bnichols@ca.afilias.info> wrote:
>
> > On Fri, 2008-04-18 at 10:38 -0700, Selena Deckelmann wrote:
> > > JD's talk and slides:
> > > http://www.commandprompt.com/blogs/joshua_drake/2008/04/what_mysql_and_really_sun_can_learn_from_postgresql/
> > >
> > > Review of the talk:
> > > http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql
> > >
> >
> > Interesting reads.
> >
> > I have to agree with the part about lack of in place upgrades.  I
> > would have say that that is the single biggest thorn in our PG using
> > side here at Afilias.  Upgrading the data layer via Slony is a
> > **massive** undertaking.
>
> The problem is not that it is inappropriate. It is that it doesn't work
> for databases of that size. Even at 500GB you can have serious issues
> with Slony.

Agreed. We're living with the pain of this with one particular support
system. We have ways to upgrade it that don't involve Slony, but it's
even more painful than the Slony upgrade process.

> This is not actually a Slony limitation but a PostgreSQL one.

Agreed.  Slony does a great job of this where it is technically
feasible.  It just a really big pain, and takes a lot of work if your
databases are anything but trivial in size.

In place upgrades is number 1 on my wish list.

Brad.

Oh Josh....

> MySQL is an embarrassment at being the “world’s most popular open source
> database”. That’s like Microsoft.

http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql


Aren't you inviting the fang bearing dolphin to leap out of the water and drag you to the bottom of the pool?




--
Robert Bernier
Open Technology Group, Inc.
One Copley Parkway, Suite 210
Morrisville, NC  27560
919-463-0999/877-258-8987
http://www.otg-nc.com

On Fri, 18 Apr 2008 16:06:46 -0400
Robert Bernier <robert.bernier5@sympatico.ca> wrote:

> Aren't you inviting the fang bearing dolphin to leap out of the water
> and drag you to the bottom of the pool?
>

I have a gun toting Mammoth who likes to eat dolphins to protect me.

http://www.commandprompt.com/images/mammoth_versus_dolphin_500.jpg

Joshua D. Drake


--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate

Joshua D. Drake wrote:

> I have a gun toting Mammoth who likes to eat dolphins to protect me.
>
> http://www.commandprompt.com/images/mammoth_versus_dolphin_500.jpg

Eric Raymond was there?  Really?

--
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

On Fri, 18 Apr 2008 16:20:11 -0400
Alvaro Herrera <alvherre@commandprompt.com> wrote:

> Joshua D. Drake wrote:
>
> > I have a gun toting Mammoth who likes to eat dolphins to protect me.
> >
> > http://www.commandprompt.com/images/mammoth_versus_dolphin_500.jpg
>
> Eric Raymond was there?  Really?
>

ouch



--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate

bnichols@ca.afilias.info (Brad Nicholson) writes:
> On Fri, 2008-04-18 at 10:38 -0700, Selena Deckelmann wrote:
>> JD's talk and slides:
>> http://www.commandprompt.com/blogs/joshua_drake/2008/04/what_mysql_and_really_sun_can_learn_from_postgresql/
>>
>> Review of the talk:
>> http://www.bytebot.net/blog/archives/2008/04/17/what-mysql-can-learn-from-postgresql
>>
>
> Interesting reads.
>
> I have to agree with the part about lack of in place upgrades.  I
> would have say that that is the single biggest thorn in our PG using
> side here at Afilias.  Upgrading the data layer via Slony is a
> **massive** undertaking.

You clearly haven't done it using eRServer ;-).

I am as "pro-having-in-place-upgrades" as it comes, but it should be
noted that upgrading via Slony-I was one of the intended use cases in
its design, and that use case has certainly been successfully
addressed.

It is by no mean obvious that we will see, any time soon, an in-place
upgrade that doesn't have big enough caveats in it[1] that you're
won't still be left with a highly non-insignificant amount of effort.

Footnotes:

[1] For instance, what if the process doesn't allow you to keep
replication 'running'?  Upgrading a cluster of 4 replicated nodes
might *still* involve redoing 3 of them, which means that nothing was
materially diminished from the former "**massive** undertaking."
--
(reverse (concatenate 'string "ofni.sesabatadxunil" "@" "enworbbc"))
http://linuxdatabases.info/info/
Black holes result from God dividing the universe by zero.

On Friday 18 April 2008 16:20, Alvaro Herrera wrote:
> Eric Raymond was there?  Really?

I met him a couple of times too at several FreeBSD conferences.


--
Robert Bernier
Open Technology Group, Inc.
One Copley Parkway, Suite 210
Morrisville, NC  27560
919-463-0999/877-258-8987
http://www.otg-nc.com

Chris Browne wrote:

> I am as "pro-having-in-place-upgrades" as it comes, but it should be
> noted that upgrading via Slony-I was one of the intended use cases in
> its design, and that use case has certainly been successfully
> addressed.

Also, plans are forthcoming for improving vacuum so that it will be no
longer such a huge problem even in "high velocity" databases.

--
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

All,

>>> I have to agree with the part about lack of in place upgrades.  I
>>> would have say that that is the single biggest thorn in our PG using
>>> side here at Afilias.  Upgrading the data layer via Slony is a
>>> **massive** undertaking.

Zdenek is working on this.  Anyone want to help?

--Josh

Josh Berkus wrote:
> All,
>
>>>> I have to agree with the part about lack of in place upgrades.  I
>>>> would have say that that is the single biggest thorn in our PG using
>>>> side here at Afilias.  Upgrading the data layer via Slony is a
>>>> **massive** undertaking.
>
> Zdenek is working on this.  Anyone want to help?

Well I think the more appropriate list would be -hackers but, I was
going to offer to help him with the linux side of testing. I will admit
though, he hasn't gotten sign off from -hackers on his implementation so
I am wary of putting any real effort into it until procedure is followed.

Joshua D. Drake

In the last exciting episode, alvherre@commandprompt.com (Alvaro Herrera) wrote:
> Chris Browne wrote:
>
>> I am as "pro-having-in-place-upgrades" as it comes, but it should be
>> noted that upgrading via Slony-I was one of the intended use cases in
>> its design, and that use case has certainly been successfully
>> addressed.
>
> Also, plans are forthcoming for improving vacuum so that it will be no
> longer such a huge problem even in "high velocity" databases.

We have seen considerable improvements to VACUUM in practically every
major release, so that between 7.2 and 8.3, it has become a "way less
huge issue."  (I don't recall there being anything *major* improved
about VACUUM in 7.3, but then, I don't remember much about 7.3, having
only barely used it.)

I'm glad to hear that the efforts have not slackened :-)!
--
(format nil "~S@~S" "cbbrowne" "gmail.com")
http://linuxfinances.info/info/postgresql.html
Never put off until tomorrow what you can put off indefinitely.

On Fri, Apr 18, 2008 at 03:34:38PM -0400, Chris Browne wrote:
> I am as "pro-having-in-place-upgrades" as it comes, but it should be
> noted that upgrading via Slony-I was one of the intended use cases in
> its design, and that use case has certainly been successfully
> addressed.

I think one needs to be careful not to overstate the success achieved.  It
does work.  That doesn't mean it's easy or quick.  Given that TheirSQL drops
the new daemon in place and upgrades the tables on the way by, upgrade via
Slony looks remarkably kludgey.

Yes, I understand the reasons why it's this way; that doesn't mean it's the
best of all possible worlds.  Dr Pangloss was wrong, durnit.

A