Skip site navigation (1) Skip section navigation (2)

Re: Fragmentation project

From: "Marko Kreen" <markokr(at)gmail(dot)com>
To: "Gustavo Tonini" <gustavotonini(at)gmail(dot)com>
Cc: "pgsql-hackers list" <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Fragmentation project
Date: 2007-04-27 11:07:43
Message-ID: (view raw, whole thread or download thread mbox)
Lists: pgsql-hackers
On 4/25/07, Gustavo Tonini <gustavotonini(at)gmail(dot)com> wrote:
> On 4/24/07, Marko Kreen <markokr(at)gmail(dot)com> wrote:
> > On 4/23/07, Heikki Linnakangas <heikki(at)enterprisedb(dot)com> wrote:
> > > Oh, you're talking about distributing partitions across different nodes
> > > and parallelizing queries. No, we don't do that today.
> >
> > PL/Proxy actually works like that, only in smaller scope -
> > for function calls only.
> >
> I think that proposed funcionalities cannot be implemented in a PL scope...

Oh, sure.  PL/proxy just proves if you can adhere to specific coding-style
- all db access goes via functions - you can solve the problem today,
with simple tools.  You dont need even PL/proxy for that, any PL that
can do connections (plpython, plperl, ..) can be used for proxy functions.
PL/proxy just makes it so much easier.

Function based DB API can give additional benefits:
- easy upgradeablility
- easy to change db structure without apps knowing
- easy to monitor/audit

And ofcourse:
- if database hits hardware limits, you can replace all functions
with PLproxy functions and spread data over several partitions.

I think most OLTP apps can be designed around db-functions, with OLAP
it will be bit harder, but there it is less critical also.


In response to

pgsql-hackers by date

Next:From: Heikki LinnakangasDate: 2007-04-27 11:22:22
Subject: Re: Avoiding unnecessary reads in recovery
Previous:From: rupesh bajajDate: 2007-04-27 11:01:48
Subject: When the locially dropped column is also physically dropped

Privacy Policy | About PostgreSQL
Copyright © 1996-2017 The PostgreSQL Global Development Group