| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Ondrej Ivanič <ondrej(dot)ivanic(at)gmail(dot)com> |
| Cc: | pgsql-general(at)postgresql(dot)org |
| Subject: | Re: Sort Method: external merge |
| Date: | 2011-08-25 14:14:05 |
| Message-ID: | 1309.1314281645@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
=?UTF-8?Q?Ondrej_Ivani=C4=8D?= <ondrej(dot)ivanic(at)gmail(dot)com> writes:
> work_mem is set to 4 000 000 kb and I do not understand why few
> queries (3 and 5) used disk and the rest fit were able to data into
> memory.
The on-disk representation of sort data is quite a bit more compact than
the in-memory representation. So where it says that 3.5GB of disk were
needed, it's not a surprise that the memory requirement would have
exceeded 4GB to do the sort in-memory.
If you want to know what the conversion factor is for your particular
dataset, try doing the query with small and large work_mem so you can
see what is reported as the amount of space needed each way.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | uanacho | 2011-08-25 15:35:33 | rollback doubt and connection to remoteDB |
| Previous Message | Samba | 2011-08-25 13:29:14 | Re: Streaming Replication: Observations, Questions and Comments |