pgbench and scaling

Started by Rakesh Kumarover 9 years ago4 messagesgeneral
Jump to latest
#1Rakesh Kumar
rakeshkumar464@outlook.com

PG 9.5.3
RHEL Linux

To cut the story short and giving as much details as I can remember:

I created 40 database db1 .. db40 in a cluster.
On each database I initialized pgbench tables with a scale of 15.

Then I concurrently ran 5 pgbenches for 15 min. Each pgbench connected to one db. In other words, no two pgbench sessions were working on the same db.

I noticed that as I scale from 5 to 10 to 20 to 40, the TPS starts falling almost linearly :

with 5, TPS was doing 639
with 10 TPS was down to 490
with 20 TPS was down to 280
and so on.

Are we missing something? Where do I start looking for the root cause.

thanks

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

#2Rakesh Kumar
rakeshkumar464@outlook.com
In reply to: Rakesh Kumar (#1)
Re: pgbench and scaling

I forgot to mention that the db is replicated synchronously. I think that is the culprit.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

#3Vick Khera
vivek@khera.org
In reply to: Rakesh Kumar (#1)
Re: pgbench and scaling

On Thu, Nov 17, 2016 at 8:08 PM, Rakesh Kumar
<rakeshkumar464@outlook.com> wrote:

I noticed that as I scale from 5 to 10 to 20 to 40, the TPS starts falling almost linearly :

with 5, TPS was doing 639
with 10 TPS was down to 490
with 20 TPS was down to 280
and so on.

Are the TPS numbers per pgbench? If so, then you're getting
10x490=4900 TPS system wide, or 20*280=5600 TPS system wide.

If those are total TPS numbers then you should check your disk
utilization and CPU utilization and see where your bottleneck is.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

#4Rakesh Kumar
rakeshkumar464@outlook.com
In reply to: Vick Khera (#3)
Re: pgbench and scaling

"Are the TPS numbers per pgbench? If so, then you're getting
10x490=4900 TPS system wide, or 20*280=5600 TPS system wide. "

Per pgbench.

Your explanation makes sense. thanks.

--
View this message in context: http://postgresql.nabble.com/pgbench-and-scaling-tp5930891p5931131.html
Sent from the PostgreSQL - general mailing list archive at Nabble.com.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general