Hi Alvaro,
On 18/08/15 17:39, Alvaro Herrera wrote:
> Joshua D. Drake wrote:
>>
>> On 08/18/2015 09:19 AM, Melvin Davidson wrote:
>>>> 8 x 16GB 1600MHz PC3-12800 DDR3 - 128GB total
>>>>> shared_buffers=60GB
>>>
>>> I would say 60GB is too high when you have 128GB system memory.
>>> Try lowering it to shared_buffers=32GB and let the O/S handle more of
>>> the work.
>>
>> I would also look at see if you are checkpointing either via segment
>> rollover or time.
>
> If it works fine under 9.2, why would it be a problem in 9.3 and 9.4?
> Keep in mind the question is not "let's make this as fast as possible"
> but rather "let's make this on par with 9.2".
My aim was simply to get 9.4 to match 9.2.13, I spent about 5 days
tweaking and re-testing many different configurations, When I initially
installed 9.2.13 (building a replacement server for our current 9.1
setup) I made two configuration changes (shared_buffers and work_mem)
and the stats were smashing 9.1.
I think I just expected the same kind of performance increase (or
atleast the same performance) when I went to 9.4.
I'm not expecting a 20% increase against 9.2.13, but I do expect to see
the same increase against 9.1.
Thanks
Michael
>
> One thing to look at is the rate of WAL generation for a set number of
> transactions. Maybe the later releases are generating more WAL due to
> multixacts, for instance (prior to 9.3 these weren't wal-logged.)
>