On Thu, Sep 24, 2015 at 9:17 PM, Ashish Samant <ashish.samant@xxxxxxxxxx> wrote:Yes. The main performance gain is due to the reduced contention on one spinlock(fc->lock) , especially with a large number of requests.
We did some performance testing without these patches and with these patchesInteresting. This means, that serving the request on a different NUMA
(with -o clone_fd option specified). We did 2 types of tests:
1. Throughput test : We did some parallel dd tests to read/write to FUSE
based database fs on a system with 8 numa nodes and 288 cpus. The
performance here is almost equal to the the per-numa patches we submitted a
while back.Please find results attached.
node as the one where the request originated doesn't appear to make
the performance much worse.