* Joshua D. Drake <jd@commandprompt.com> wrote:
<snip>
> There is only one production and shipping replication that does not
> require triggers that I know of and that is ours (Command Prompt).
> It is also not a userland app but actually part of the PostgreSQL engine.
Ah. Has pgsql to be patched for that or can it be simply dropped-in
like external (C) functions/procedures ?
What about lazy replication (i.e. when some node is running but not
online for a while), masterless replication (all nodes are equal)
or journaling ?
My application/framework-level solves this, but
a) requires some special table layout (first attrs have to be inode+mtime),
b) not suited for high update-rates (relly too slow for that)
but it is already working quite stable for several years :)
<snip>
> Each solution has pro's and cons. Slony-I for example appears to be
> better when doing mass updates or deletes than Replicator.
yeah, on subsequent mass-updates on just one host per table at certain time,
my stuff is also good, since it works w/ batching. but when multiple hosts
do heavy updates on the same table, it may run into conflicts.
<snip>
> On the argument of significant features for 7.5:
>
> Win32 Native
> PITR
Stupid question: whats PITR ?
> Nested Transactions
> Background Writer....
Background-Write = write-behind ?
BTW: how stable is the current pgsql against powerfails or kill -9 ?
cu
--
---------------------------------------------------------------------
Enrico Weigelt == metux IT service
phone: +49 36207 519931 www: http://www.metux.de/
fax: +49 36207 519932 email: contact@metux.de
cellphone: +49 174 7066481
---------------------------------------------------------------------
-- DSL ab 0 Euro. -- statische IP -- UUCP -- Hosting -- Webshops --
---------------------------------------------------------------------
