From: | Achilleas Mantzios <achill(at)matrix(dot)gatewaynet(dot)com> |
---|---|
To: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Memory usage per session |
Date: | 2016-07-08 14:39:46 |
Message-ID: | 577FBB32.5030306@matrix.gatewaynet.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On 08/07/2016 17:07, AMatveev(at)bitec(dot)ru wrote:
> Hi
>
>>>>> The test performs about 11K lines of code
>>>>> Oracle: about 5M
>>>>> postgreSql: about 160М
>
>> Do you have 100 CPUs on this system which apparently doesn't have 16G
>> of RAM available for PG to use?
> We can say at fact:
> We currently work at oracle.
> Our code base about 4000 k line of code
> In out last project we have:
> 3000 current connection
> 200 active session
> So 16g it's very optimistic.
> Of course we think about buy hardware or software.
> It's other question.
> So with this memory consumption it can be really cheaper to by Oracle.
>> If not, you should probably consider connection pooling to reduce the
>> number of PG sessions to something approaching the number of CPUs/cores
>> you have in the system.
> It's possible only with application server,
No, you can deploy PgPool or PgBouncer.
Apart from that, I just checked in my system. User sessions have size of 16M. Not 160M.
> for local network thick client has reasonable advantages.
> We just can't implement today all functions on thin client.
>
>
>
--
Achilleas Mantzios
IT DEV Lead
IT DEPT
Dynacom Tankers Mgmt
From | Date | Subject | |
---|---|---|---|
Next Message | AMatveev | 2016-07-08 15:32:25 | Re: Memory usage per session |
Previous Message | Tom Lane | 2016-07-08 14:36:48 | Re: pasting a lot of commands to psql |