On Wed, 23 Sep 2015, Deneau, Tom wrote: > Hi all -- > > Looking for guidance with perf counters... > I am trying to see whether the perf counters can tell me anything about the following discrepancy > > I populate a number of 40k size objects in each of two pools, poolA and poolB. > Both pools cover osds on a single node, 5 osds total. > > * Config 1 (1p): > * use single rados bench client with 32 threads to do seq read of 20000 objects from poolA. > > * Config 2 (2p): > * use two concurrent rados bench clients (running on same client node) with 16 threads each, > one reading 10000 objects from poolA, > one reading 10000 objects from poolB, > > So in both configs, we have 32 threads total and the number of objects read is the same. > Note: in all cases, we drop the caches before doing the seq reads > > The combined bandwidth (MB/sec) for the 2 clients in config 2 is about 1/3 of the bandwidth for > the single client in config 1. How were the object written? I assume the cluster is backed by spinning disks? I wonder if this is a disk layout issue. If the 20,000 objects are written in order, they willb e roughly sequential on disk, and the 32 thread case will read them in order. In the 2x 10,000 case, the two clients are reading two sequences of objects written at different times, and the disk arms will be swinging around more. My guess is that if the reads were reading the objects in a random order the performance would be the same... I'm not sure that rados bench does that though? sage > > > I gathered perf counters before and after each run and looked at the difference of > the before and after counters for both the 1p and 2p cases. Here are some things I noticed > that are different between the two runs. Can someone take a look and let me know > whether any of these differences are significant. In particular, for the > throttle-msgr_dispatch_throttler ones, since I don't know the detailed definitions of these fields. > Note: these are the numbers for one of the 5 osds, the other osds are similar... > > * The field osd/loadavg is always about 3 times higher on the 2p c > > some latency-related counters > ------------------------------ > osd/op_latency/sum 1p=6.24801117205061, 2p=579.722513078945 > osd/op_process_latency/sum 1p=3.48506945394911, 2p=42.6278494549915 > osd/op_r_latency/sum 1p=6.2480111719924, 2p=579.722513079003 > osd/op_r_process_latency/sum 1p=3.48506945399276, 2p=42.6278494550061 > > > and some throttle-msgr_dispatch_throttler related counters > ---------------------------------------------------------- > throttle-msgr_dispatch_throttler-client/get 1p=1337, 2p=1339, diff=2 > throttle-msgr_dispatch_throttler-client/get_sum 1p=222877, 2p=223088, diff=211 > throttle-msgr_dispatch_throttler-client/put 1p=1337, 2p=1339, diff=2 > throttle-msgr_dispatch_throttler-client/put_sum 1p=222877, 2p=223088, diff=211 > throttle-msgr_dispatch_throttler-hb_back_server/get 1p=58, 2p=134, diff=76 > throttle-msgr_dispatch_throttler-hb_back_server/get_sum 1p=2726, 2p=6298, diff=3572 > throttle-msgr_dispatch_throttler-hb_back_server/put 1p=58, 2p=134, diff=76 > throttle-msgr_dispatch_throttler-hb_back_server/put_sum 1p=2726, 2p=6298, diff=3572 > throttle-msgr_dispatch_throttler-hb_front_server/get 1p=58, 2p=134, diff=76 > throttle-msgr_dispatch_throttler-hb_front_server/get_sum 1p=2726, 2p=6298, diff=3572 > throttle-msgr_dispatch_throttler-hb_front_server/put 1p=58, 2p=134, diff=76 > throttle-msgr_dispatch_throttler-hb_front_server/put_sum 1p=2726, 2p=6298, diff=3572 > throttle-msgr_dispatch_throttler-hbclient/get 1p=168, 2p=252, diff=84 > throttle-msgr_dispatch_throttler-hbclient/get_sum 1p=7896, 2p=11844, diff=3948 > throttle-msgr_dispatch_throttler-hbclient/put 1p=168, 2p=252, diff=84 > throttle-msgr_dispatch_throttler-hbclient/put_sum 1p=7896, 2p=11844, diff=3948 > > -- Tom Deneau, AMD > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
