Re: Will higher shared_buffers improve tpcb-like benchmarks?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Please remove me from this list Serv.  I do not use this db anymore and fills my alerts daily.  


On Tue, Jan 29, 2019 at 06:39 Saurabh Nanda <saurabhnanda@xxxxxxxxx> wrote:
Hi,

I'm going crazy trying to optimise my Postgres config for a production setting [1] Once I realised random changes weren't getting my anywhere, I finally purchased PostgreSQL 10 - Higher Performance [2] and understood the impact of shared_buffers.

IIUC, shared_buffers won't have any significant impact in the following scenario, right?

-- DB size = 30GB
-- shared_buffers = 2GB
-- workload = tpcb-like

This is because the tpcb-like workload selects & updates random rows from the DB [3]. Therefore, with a 2GB shared buffer, there is only a 6-7% chance (did I get my probability correct?) that the required data will be in the shared_buffer. Did I understand this correctly?

If nothing else becomes the bottleneck (eg. periodically writing dirty pages to disk), increasing the shared_buffers to 15GB+ should have a significant impact, for this DB-size and workload, right? (The system has 64 GB RAM)


-- Saurabh.
--
Ethical axioms are found and tested not very differently from the axioms of science. Truth is what stands the the test if experience.

[Postgresql General]     [Postgresql PHP]     [PHP Users]     [PHP Home]     [PHP on Windows]     [Kernel Newbies]     [PHP Classes]     [PHP Books]     [PHP Databases]     [Yosemite]

  Powered by Linux