From: | Joshua Drake <jd(at)commandprompt(dot)com> |
---|---|
To: | Don Seiler <don(at)seiler(dot)us> |
Cc: | pgsql-performance(at)lists(dot)postgresql(dot)org |
Subject: | Re: High COMMIT times |
Date: | 2021-01-06 16:51:06 |
Message-ID: | CAJvJg-SUfe+Fv0JmvBWgyaPx=J9dzEk9Bg=4JheuVwAdNMj6HA@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
> I have the stats_temp_directory in a tmpfs mount. I *do* have pg_wal on
> the same premium SSD storage volume as the data directory. Normally I would
> know to separate these but I was told with the cloud storage that it's all
> virtualized anyway, plus storage IOPS are determined by disk size so having
> a smaller volume just for pg_wal would hurt me in this case. The kind folks
> in the PG community Slack suggested just having one large premium cloud
> storage mount for
>
the data directory and leave pg_wal inside because this virtualization
> removes any guarantee of true separation.
>
It is true that the IO is virtualized but that does not mean that separate
volumes won't help. In cloud storage you are granted specific IOPS/MB/s per
volume. Separating pg_wal to a new volume mount will take pressure off of
page writes and allow the wal to write within its own prioritization.
JD
From | Date | Subject | |
---|---|---|---|
Next Message | Don Seiler | 2021-01-06 18:06:27 | Re: High COMMIT times |
Previous Message | Don Seiler | 2021-01-06 16:19:19 | High COMMIT times |