After 2.6.30 I am seeing large performance regressions on a raid setup.
I am working to publish a larger amount of data but I wanted to get some
quick data out about what I am seeing.
The test (FFSB test suite) I am running is basically random direct io
writes. The below data is from 128 threads all doing these random
writes. 1 and 32 thread results are not as drastically bad but 2.6.30
has the strongest results.
Under a mailserver workload I see similar performance impacts at this
same kernel change point. I hope to publish better data soon. Several
other workload types do not show this performance regression.
2.6.30:
Total Results
===============
Op Name Transactions Trans/sec % Trans % Op Weight
Throughput
======= ============ ========= ======= ===========
==========
write : 9015040 29561.46 100.000% 100.000%
115MB/sec
-
29561.46 Transactions per Second
Any kernel past 2.6.30. This is from 2.6.31-rc1:
Total Results
===============
Op Name Transactions Trans/sec % Trans % Op Weight
Throughput
======= ============ ========= ======= ===========
==========
write : 3185920 10120.50 100.000% 100.000%
39.5MB/sec
-
10120.50 Transactions per Second
Nothing is sticking out in oprofile data with this workload.
I am in the middle of a kernel git bisect that should be done in another
day or so.
Thanks,
Keith Mannthey
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
