pg.ddx.io pgsql-admin@postgresql.org mailing list archive
help / color / mirror / Atom feedBetter I/O throughput? (was Re: create tablespace - cannot run inside a transaction block)
3+ messages / 3 participants
[nested] [flat]
* Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block)
@ 2019-09-25 19:40 Ron <ronljohnsonjr@gmail.com>
2019-09-25 19:44 ` Re: Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block) Avin Kavish <avinkavish@gmail.com>
2019-09-26 18:57 ` Re: Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block) Laurenz Albe <laurenz.albe@cybertec.at>
0 siblings, 2 replies; 3+ messages in thread
From: Ron @ 2019-09-25 19:40 UTC (permalink / raw)
To: pgsql-admin@lists.postgresql.org
On 9/25/19 2:16 PM, Laurenz Albe wrote:
> On Wed, 2019-09-25 at 14:50 +0000, Pepe TD Vo wrote:
>> Normally, in Oracle we need to create database, tablespace then
>> username/schema and tables, objects, etc...
>>
>> is the procedure as same as in Postgres? I see the login and schema
>> are totally different in Postgres.
> No, normally you don't create tablespaces in PostgreSQL.
> They are a few use cases for them, but not many.
Do I/O requests in the Linux kernel get "backlogged" when they all hit the
same device? Or would you get better throughput (or less latency) by
spreading the load across multiple devices?
(A long-running synchronous IO request seems it would block everything
behind it, whereas objects in tablespaces on different devices could still
be queried.)
--
Angular momentum makes the world go 'round.
^ permalink raw reply [nested|flat] 3+ messages in thread
* Re: Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block)
2019-09-25 19:40 Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block) Ron <ronljohnsonjr@gmail.com>
@ 2019-09-25 19:44 ` Avin Kavish <avinkavish@gmail.com>
1 sibling, 0 replies; 3+ messages in thread
From: Avin Kavish @ 2019-09-25 19:44 UTC (permalink / raw)
To: Ron <ronljohnsonjr@gmail.com>; +Cc: pgsql-admin@lists.postgresql.org
In reply to the original question being raised about an RDS instance,
afaik, there's no need to do tablespaces on RDS as IOPS is provisioned as
requested, the actual hardware implementation is abstracted away and
irrelevant.
On Thu., 26 Sep. 2019, 1:10 am Ron, <ronljohnsonjr@gmail.com> wrote:
> On 9/25/19 2:16 PM, Laurenz Albe wrote:
> > On Wed, 2019-09-25 at 14:50 +0000, Pepe TD Vo wrote:
> >> Normally, in Oracle we need to create database, tablespace then
> >> username/schema and tables, objects, etc...
> >>
> >> is the procedure as same as in Postgres? I see the login and schema
> >> are totally different in Postgres.
> > No, normally you don't create tablespaces in PostgreSQL.
> > They are a few use cases for them, but not many.
>
> Do I/O requests in the Linux kernel get "backlogged" when they all hit the
> same device? Or would you get better throughput (or less latency) by
> spreading the load across multiple devices?
>
> (A long-running synchronous IO request seems it would block everything
> behind it, whereas objects in tablespaces on different devices could still
> be queried.)
>
> --
> Angular momentum makes the world go 'round.
>
>
>
^ permalink raw reply [nested|flat] 3+ messages in thread
* Re: Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block)
2019-09-25 19:40 Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block) Ron <ronljohnsonjr@gmail.com>
@ 2019-09-26 18:57 ` Laurenz Albe <laurenz.albe@cybertec.at>
1 sibling, 0 replies; 3+ messages in thread
From: Laurenz Albe @ 2019-09-26 18:57 UTC (permalink / raw)
To: Ron <ronljohnsonjr@gmail.com>; pgsql-admin@lists.postgresql.org
On Wed, 2019-09-25 at 14:40 -0500, Ron wrote:
> > No, normally you don't create tablespaces in PostgreSQL.
> > They are a few use cases for them, but not many.
>
> Do I/O requests in the Linux kernel get "backlogged" when they all
> hit the
> same device? Or would you get better throughput (or less latency)
> by
> spreading the load across multiple devices?
Sure, spreading I/O across multiple devices is one use case for
tablespaces. But usually this is easier to do by striping a
logical volume across several physical volumes at a lower level.
Yours,
Laurenz Albe
--
Cybertec | https://www.cybertec-postgresql.com
^ permalink raw reply [nested|flat] 3+ messages in thread
end of thread, other threads:[~2019-09-26 18:57 UTC | newest]
Thread overview: 3+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2019-09-25 19:40 Better I/O throughput? (was Re: create tablespace - cannot run inside a transaction block) Ron <ronljohnsonjr@gmail.com>
2019-09-25 19:44 ` Avin Kavish <avinkavish@gmail.com>
2019-09-26 18:57 ` Laurenz Albe <laurenz.albe@cybertec.at>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox