Hi Mike
Thanks verrry much for your work and test cases.
From: Xiubo Li <lixiubo@xxxxxxxxxxxxxxxxxxxx>
Currently for the TCMU, the ring buffer size is fixed to 64K cmd
area + 1M data area, and this will be bottlenecks for high iops.
Hi Xiubo, thanks for your work.
daynmic -> dynamic
Have you benchmarked this patch and determined what kind of iops
improvement it allows? Do you see the data area reaching its
fully-allocated size?
I tested this patch with Venky's tcmu-runner rbd aio patches, with one
10 gig iscsi session, and for pretty basic fio direct io (64 -256K
read/writes with a queue depth of 64 numjobs between 1 and 4) tests read
throughput goes from about 80 to 500 MB/s.
Looks nice.
Write throughput is pretty
low at around 150 MB/s.
What's the original write throughput without this patch? Is it also
around 80 MB/s ?
I did not hit the fully allocated size. I did not drive a lot of IO though.
Yes, if the cmd area won't hit the fully allocated size, the data area is
also very hard to hit the fully allocated size.
And for the 64MB cmd area size is a little larger for 1GB data area.
Next, I will down it to smaller as needed.
Thanks,
BRs
Xiubo
