Re: Benchmarking: How to identify bottleneck (limiting factor) and achieve "linear scalability"?

>
> smartctl is a good start
>
Here's the output of `smartctl --xall /dev/sda` --
https://gist.github.com/saurabhnanda/ec3c95c1eb3896b3efe55181e7c78dde

I've disabled RAID so /dev/sda is the only disk which is being currently
used.

I'm still seeing very weird numbers. There seems to be absolutely no
difference in performance if I increase shared_buffers from 8MB to 2GB. Is
there some other setting which needs to be changed to take advantage of the
increase in shared_buffers? Also, something even weirder is happening for
client=1.

Is my benchmarking script alright -
https://gist.github.com/saurabhnanda/78e8523cf079ce7b78704acb1aa8d9fc ?

+--------+--------------+----------------+
| client | Defaults [1] | buffers=2G [2] |
+--------+--------------+----------------+
| 1 | 348-475 (??) | 529-583 (??) |
+--------+--------------+----------------+
| 4 | 436-452 | 451-452 |
+--------+--------------+----------------+
| 8 | 862-869 | 859-861 |
+--------+--------------+----------------+
| 12 | 1210-1219 | 1220-1225 |
+--------+--------------+----------------+

[1] Default settings
checkpoint_completion_target=0.5
default_statistics_target=100
effective_io_concurrency=1
max_parallel_workers=8
max_parallel_workers_per_gather=2
max_wal_size=1024 MB
max_worker_processes=20
min_wal_size=80 MB
random_page_cost=4
* shared_buffers=1024 8kB*
wal_buffers=32 8kB
work_mem=4096 kB

[2] Increased shared_buffers
checkpoint_completion_target=0.5
default_statistics_target=100
effective_io_concurrency=1
max_parallel_workers=8
max_parallel_workers_per_gather=2
max_wal_size=1024 MB
max_worker_processes=20
min_wal_size=80 MB
random_page_cost=4
* shared_buffers=262144 8kB*
wal_buffers=2048 8kB
work_mem=4096 kB

-- Saurabh.