Re: splitting up tables based on read/write frequency of columns - Mailing list pgsql-general

From Jonathan Vanasco
Subject Re: splitting up tables based on read/write frequency of columns
Date
Msg-id D0D8DE09-9A28-4F43-B014-6F04350D2329@2xlp.com
Whole thread Raw
In response to Re: splitting up tables based on read/write frequency of columns  (Stefan Keller <sfkeller@gmail.com>)
List pgsql-general
On Jan 19, 2015, at 5:07 PM, Stefan Keller wrote:

> Hi
>
> I'm pretty sure PostgreSQL can handle this.
> But since you asked with a theoretic background,
> it's probably worthwhile to look at column stores (like [1]).


Wow. I didn't know there was a column store extension for PG -- this would come in handy for some analytic stuff we
run!

I know that PG can handle my current system "at scale".  I'm really just wondering what the possible
slowdowns/improvementswill be.   

Doing a rewrite of the entire row + updating the various indexes seems to be a lot of unnecessary IO.  At some point it
willmake sense to minimize that and isolate the heavy-write columns from impacting the rest of the table's performance. 

pgsql-general by date:

Previous
From: Pierre Hsieh
Date:
Subject: Fwd: Ask for a question
Next
From: Raymond O'Donnell
Date:
Subject: Re: Fwd: Ask for a question