Re: Scalability

Поиск
Список
Период
Сортировка
От volunteer@spatiallink.org
Тема Re: Scalability
Дата
Msg-id 20061029072912.b22b5ede89d48a4249261b5ab56693f4.23ac476110.wbe@email.secureserver.net
обсуждение исходный текст
Ответ на Scalability  ("Bill" <postgresql@dbginc.com>)
Ответы Re: Scalability  ("Joshua D. Drake" <jd@commandprompt.com>)
Список pgsql-general
Top at 350-400 concurrent connections! Although more than my project's
peak estimates, it is a bit discomforting. Are there any *promising*
load testing numbers with Windows 2003? I'd be happy to share results
from my simulations.

Matt

> -------- Original Message --------
> Subject: Re: [GENERAL] Scalability
> From: "Joshua D. Drake" <jd@commandprompt.com>
> Date: Sat, October 28, 2006 8:38 pm
> To: Bill <postgresql@dbginc.com>
> Cc: pgsql-general@postgresql.org
>
> Bill wrote:
> > I am new to PostgreSQL and just beginning to learn the product. I will
> > probrobably be using it exclusively on Windows.
> >
> > I was surprised to learn that PostgreSQL creates a new process for each
> > connection. Doesn't this severely limit its scalability by consuming
> > resources rapidly on the server as the number of user increases?
>
> The Windows version is not anywhere near as scalable as the unix
> versions. Depending on your hardware you will top out a Windows
> installation about about 350-400 connections. You can get more out of
> Windows by modifying the registry but I am unsure of how far it will go.
>
> I have Linux installations that happily hum along with 2000-5000
> connections.
>
> So in answer to your question, in general -- no the process methodology
> we use does not limit scalability and it makes our code base much
> simpler that the  equivalent threading model.
>
> Sincerely,
>
> Joshua D. Drake


В списке pgsql-general по дате отправления:

Предыдущее
От: Robert Treat
Дата:
Сообщение: Re: Replicating changes
Следующее
От: Andrew Sullivan
Дата:
Сообщение: Re: Replicating changes