clusterperf September 13, 2011
- John Ousterhout
Owned by John Ousterhout
clusterperf output as measured on September 13, 2011:
basic.read100 5.3 us read single 100B object
basic.readBw100 18.1 MB/s bandwidth reading 100B object
basic.read1K 7.1 us read single 1KB object
basic.readBw1K 134.5 MB/s bandwidth reading 1KB object
basic.read10K 10.2 us read single 10KB object
basic.readBw10K 936.1 MB/s bandwidth reading 10KB object
basic.read100K 44.9 us read single 100KB object
basic.readBw100K 2.1 GB/s bandwidth reading 100KB object
basic.read1M 427.8 us read single 1MB object
basic.readBw1M 2.2 GB/s bandwidth reading 1MB object
basic.write100 15.0 us write single 100B object
basic.writeBw100 6.3 MB/s bandwidth writing 100B object
basic.write1K 19.4 us write single 1KB object
basic.writeBw1K 49.3 MB/s bandwidth writing 1KB object
basic.write10K 38.3 us write single 10KB object
basic.writeBw10K 248.8 MB/s bandwidth writing 10KB object
basic.write100K 221.4 us write single 100KB object
basic.writeBw100K 430.8 MB/s bandwidth writing 100KB object
basic.write1M 2.2 ms write single 1MB object
basic.writeBw1M 436.9 MB/s bandwidth writing 1MB object
broadcast 181.1 us broadcast message to 9 slaves
netBandwidth 110.2 GB/s many clients reading from different servers
netBandwidth.max 1.6 GB/s fastest client
netBandwidth.min 470.4 MB/s slowest client
readNotFound 12.4 us read object that doesn't exist
# RAMCloud read performance as a function of load (1 or more
# clients all reading a single 100-byte object repeatedly).
# Generated by 'clusterperf.py readLoaded'
#
# numClients readLatency(us) throughput(total kreads/sec)
#----------------------------------------------------------
1 5.2 193
2 6.0 331
3 5.5 541
4 6.1 650
5 6.6 758
6 7.1 841
7 8.4 834
8 9.4 847
9 10.8 833
10 12.0 834
11 13.2 835
12 14.4 834
13 15.6 831
14 16.9 828
15 18.5 811
16 19.4 823
17 20.7 822
18 21.8 825
19 22.8 832
20 24.2 825
# RAMCloud read performance when 1 or more clients read
# 100-byte objects chosen at random from 10 servers.
# Generated by 'clusterperf.py readRandom'
#
# numClients throughput(total kreads/sec) slowest(ms) reads > 10us
#--------------------------------------------------------------------
1 182 0.09 0.1%
2 342 0.33 0.1%
3 520 0.28 0.2%
4 674 2.05 0.2%
5 858 0.37 0.2%
6 1021 0.34 0.2%
7 1178 0.34 0.2%
8 1358 0.34 0.2%
9 1506 0.35 0.2%
10 1668 0.33 0.2%
11 1800 0.34 0.3%
12 1936 0.59 0.3%
13 2088 0.50 0.4%
14 2267 0.35 0.3%
15 2375 0.35 0.4%
16 2515 0.36 0.4%
17 2645 0.35 0.5%
18 2804 0.34 0.5%
19 2929 0.36 0.6%
20 3049 0.35 0.7%
21 3175 0.35 0.8%
22 3280 0.49 0.9%
23 3433 0.34 1.0%
24 3532 0.35 1.3%
25 3645 0.41 1.5%
26 3782 0.35 1.4%
27 3880 0.36 1.8%
28 4044 0.36 2.0%
29 4139 0.35 2.0%
30 4244 0.36 2.3%
31 4318 0.47 2.9%
32 4483 0.36 3.0%
33 4587 0.35 3.3%
34 4661 1.94 3.9%
35 4723 2.22 4.4%
36 4841 2.19 4.5%
37 4931 2.26 4.7%
38 5068 2.28 5.8%
39 5123 2.16 6.3%
40 5210 2.07 7.0%
41 5252 2.27 7.9%
42 5382 2.28 8.4%
43 5485 2.32 8.8%
44 5507 2.31 10.1%
45 5511 2.32 11.4%
46 5528 3.57 11.1%
47 5695 3.71 12.1%
48 5809 2.33 13.4%
49 5686 2.54 13.6%
50 5869 2.47 14.3%