Running a benchmark :
rados bench -p cephfs_data 300 write --no-cleanup

While watching ceph client io:
ceph -w

I get different numbers.

The output form rados bench is as follows:
Total time run:         300.108725
Total writes made:      66306
Write size:             4194304
Bandwidth (MB/sec):     883.760

Stddev Bandwidth:       65.2804
Max bandwidth (MB/sec): 968
Min bandwidth (MB/sec): 0
Average Latency:        0.0724089
Stddev Latency:         0.0415512
Max latency:            0.674297
Min latency:            0.0299922

While watching the client io it shoots up to around 1900 MB/s:
2015-05-18 16:17:16.416059 mon.0 [INF] pgmap v774: 4096 pgs: 4096
active+clean; 203 GB data, 607 GB used, 173 TB / 174 TB avail; 1870 MB/s
wr, 467 op/s

The max bandwidth that rados reported was 968 MB/s.

The pool is rep 3 with 48 OSDs 4096 PGs.

Any thoughts?
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to