| From: | Krzysztof Chodak <krzysztof(dot)chodak(at)gmail(dot)com> |
|---|---|
| To: | Merlin Moncure <mmoncure(at)gmail(dot)com>, andy(at)squeakycode(dot)net |
| Cc: | pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: Variable versus constrant size tuples |
| Date: | 2011-08-19 15:19:34 |
| Message-ID: | CAMiSJ3pgqPE851Ra3Ri3R6vkD8BODamQJ3R9tNWjnkA6oT0EJA@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
Thank you. Now I see that page consists of record pointers list build
from offset and length so there is no benefit of having constant
length here.
On Fri, Aug 19, 2011 at 17:03, Merlin Moncure <mmoncure(at)gmail(dot)com> wrote:
> On Fri, Aug 19, 2011 at 4:03 AM, Krzysztof Chodak
> <krzysztof(dot)chodak(at)gmail(dot)com> wrote:
>> Is there any performance benefit of using constant size tuples?
>
> not really. If your tuple size is under a known maximum length, then
> a toast table doesn't have to be created. that's a pretty minor
> detail though.
>
> merlin
>
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Greg Smith | 2011-08-19 17:20:27 | Re: Reports from SSD purgatory |
| Previous Message | Merlin Moncure | 2011-08-19 15:03:52 | Re: Variable versus constrant size tuples |