From: | Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com> |
---|---|
To: | Suvankar Roy <suvankar(dot)roy(at)tcs(dot)com> |
Cc: | Alex Goncharov <alex-goncharov(at)comcast(dot)net>, pgsql-performance(at)postgresql(dot)org |
Subject: | Re: Performance comparison between Postgres and Greenplum |
Date: | 2009-07-17 06:24:35 |
Message-ID: | dcc563d10907162324o142f1f35r8f9b3c61d6a78203@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
On Wed, Jul 15, 2009 at 7:02 AM, Suvankar Roy<suvankar(dot)roy(at)tcs(dot)com> wrote:
>
> Hi Alex,
>
> Yes, I have got 2 segments and a master host. So, in a way processing should
> be faster in Greenplum.
>
> Actually this is only a sort of Proof of Concept trial that I am carrying
> out to notice differences between greenplum and postgres, if any.
You're definitely gonna want more data to test with. I run regular
vanilla pgsql for stats at work, and we average 0.8M to 2M rows of
stats every day. We keep them for up to two years. So, when we reach
our max of two years, we're talking somewhere in the range of a
billion rows to mess about with.
During a not so busy day, the 99,000th row entered into stats for
happens at about 3am. Once they're loaded into memory it takes 435 ms
to access those 99k rows.
Start testing in the millions, at a minimum. Hundreds of millions is
more likely to start showing a difference.
From | Date | Subject | |
---|---|---|---|
Next Message | Scara Maccai | 2009-07-17 06:45:18 | Re: cluster index on a table |
Previous Message | Scott Carey | 2009-07-17 03:07:43 | Re: cluster index on a table |