clusterperf August 8, 2014
- John Ousterhout
clusterperf output measured on August 8, 2014 using rc02-rc20.
Note: these measurements were made with "-T tcp-1g" (it used TcpTransport over 1GigE). Thus, these results should provide a reasonable comparison point for people outside Stanford trying RAMCloud on traditional 1GigE networks.
basic.read100 87.3 us read single 100B object (30B key) median
basic.read100.min 50.0 us read single 100B object (30B key) minimum
basic.read100.9 99.8 us read single 100B object (30B key) 90%
basic.read100.99 144.7 us read single 100B object (30B key) 99%
basic.read100.999 180.8 us read single 100B object (30B key) 99.9%
basic.readBw100 1.2 MB/s bandwidth reading 100B object (30B key)
basic.read1K 97.2 us read single 1KB object (30B key) median
basic.read1K.min 80.4 us read single 1KB object (30B key) minimum
basic.read1K.9 140.4 us read single 1KB object (30B key) 90%
basic.read1K.99 150.1 us read single 1KB object (30B key) 99%
basic.read1K.999 188.4 us read single 1KB object (30B key) 99.9%
basic.readBw1K 9.4 MB/s bandwidth reading 1KB object (30B key)
basic.read10K 350.0 us read single 10KB object (30B key) median
basic.read10K.min 304.3 us read single 10KB object (30B key) minimum
basic.read10K.9 433.2 us read single 10KB object (30B key) 90%
basic.read10K.99 501.2 us read single 10KB object (30B key) 99%
basic.read10K.999 639.6 us read single 10KB object (30B key) 99.9%
basic.readBw10K 25.3 MB/s bandwidth reading 10KB object (30B key)
basic.read100K 1.4 ms read single 100KB object (30B key) median
basic.read100K.min 1.1 ms read single 100KB object (30B key) minimum
basic.read100K.9 1.4 ms read single 100KB object (30B key) 90%
basic.read100K.99 1.5 ms read single 100KB object (30B key) 99%
basic.read100K.999 1.6 ms read single 100KB object (30B key) 99.9%
basic.readBw100K 70.9 MB/s bandwidth reading 100KB object (30B key)
basic.read1M 8.6 ms read single 1MB object (30B key) median
basic.read1M.min 8.4 ms read single 1MB object (30B key) minimum
basic.read1M.9 8.7 ms read single 1MB object (30B key) 90%
basic.read1M.99 8.8 ms read single 1MB object (30B key) 99%
basic.readBw1M 111.2 MB/s bandwidth reading 1MB object (30B key)
basic.write100 219.4 us write single 100B object (30B key) median
basic.write100.min 151.1 us write single 100B object (30B key) minimum
basic.write100.9 290.2 us write single 100B object (30B key) 90%
basic.write100.99 321.6 us write single 100B object (30B key) 99%
basic.write100.999 1.7 ms write single 100B object (30B key) 99.9%
basic.writeBw100 413.3 KB/s bandwidth writing 100B object (30B key)
basic.write1K 308.7 us write single 1KB object (30B key) median
basic.write1K.min 220.6 us write single 1KB object (30B key) minimum
basic.write1K.9 340.5 us write single 1KB object (30B key) 90%
basic.write1K.99 404.0 us write single 1KB object (30B key) 99%
basic.write1K.999 1.6 ms write single 1KB object (30B key) 99.9%
basic.writeBw1K 3.1 MB/s bandwidth writing 1KB object (30B key)
basic.write10K 1.0 ms write single 10KB object (30B key) median
basic.write10K.min 861.5 us write single 10KB object (30B key) minimum
basic.write10K.9 1.2 ms write single 10KB object (30B key) 90%
basic.write10K.99 1.4 ms write single 10KB object (30B key) 99%
basic.write10K.999 3.8 ms write single 10KB object (30B key) 99.9%
basic.writeBw10K 8.7 MB/s bandwidth writing 10KB object (30B key)
basic.write100K 4.6 ms write single 100KB object (30B key) median
basic.write100K.min 4.4 ms write single 100KB object (30B key) minimum
basic.write100K.9 4.8 ms write single 100KB object (30B key) 90%
basic.write100K.99 5.6 ms write single 100KB object (30B key) 99%
basic.writeBw100K 20.6 MB/s bandwidth writing 100KB object (30B key)
basic.write1M 34.2 ms write single 1MB object (30B key) median
basic.write1M.min 34.0 ms write single 1MB object (30B key) minimum
basic.write1M.9 35.1 ms write single 1MB object (30B key) 90%
basic.write1M.99 35.4 ms write single 1MB object (30B key) 99%
basic.writeBw1M 27.8 MB/s bandwidth writing 1MB object (30B key)
# RAMCloud multiRead performance for 100 B objects with 30 byte keys
# located on a single master.
# Generated by 'clusterperf.py multiRead_oneMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 85.1 85.15
2 1 2 100.2 50.10
3 1 3 100.7 33.56
4 1 4 111.4 27.85
5 1 5 111.9 22.37
6 1 6 108.0 18.00
7 1 7 113.5 16.22
8 1 8 118.8 14.85
9 1 9 113.9 12.66
10 1 10 138.1 13.81
20 1 20 189.1 9.45
30 1 30 288.5 9.62
40 1 40 607.1 15.18
50 1 50 40607.9 812.16
60 1 60 40308.6 671.81
70 1 70 40503.8 578.63
80 1 80 40953.8 511.92
90 1 90 41058.0 456.20
100 1 100 41001.4 410.01
200 1 200 2036.6 10.18
300 1 300 1832.0 6.11
400 1 400 1741.8 4.35
500 1 500 1734.0 3.47
600 1 600 1918.2 3.20
700 1 700 1737.0 2.48
800 1 800 20073.2 25.09
900 1 900 2491.7 2.77
1000 1 1000 2898.7 2.90
2000 1 2000 14551.3 7.28
3000 1 3000 5874.4 1.96
4000 1 4000 10100.5 2.53
5000 1 5000 12620.2 2.52
# RAMCloud multiRead performance for 100 B objects with 30 byte keys
# with one object located on each master.
# Generated by 'clusterperf.py multiRead_oneObjectPerMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 101.3 101.28
2 2 1 114.6 57.28
3 3 1 145.2 48.39
4 4 1 156.5 39.13
5 5 1 155.4 31.07
6 6 1 153.2 25.53
7 7 1 152.6 21.80
8 8 1 170.9 21.36
9 9 1 188.2 20.91
10 10 1 164.8 16.48
11 11 1 164.0 14.91
12 12 1 207.2 17.27
13 13 1 233.0 17.92
14 14 1 298.9 21.35
15 15 1 295.3 19.69
16 16 1 288.6 18.04
17 17 1 281.4 16.55
18 18 1 320.1 17.78
# RAMCloud multi-read throughput of a single server with a
# varying number of clients issuing 70-object multi-reads on
# randomly-chosen 100-byte objects with 30-byte keys
# Generated by 'clusterperf.py multiReadThroughput'
#
# numClients throughput worker utiliz.
# (kreads/sec)
#-------------------------------------------
1 2 0.001
2 4 0.002
3 5 0.002
4 7 0.003
5 9 0.004
6 10 0.005
7 12 0.005
8 14 0.006
9 16 0.007
10 18 0.007
11 19 0.008
12 21 0.009
13 23 0.010
# RAMCloud multiWrite performance for 100 B objects with 30 byte keys
# located on a single master.
# Generated by 'clusterperf.py multiWrite_oneMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 242.4 242.37
2 1 2 287.2 143.59
3 1 3 299.4 99.81
4 1 4 351.3 87.82
5 1 5 305.6 61.11
6 1 6 330.6 55.09
7 1 7 345.1 49.30
8 1 8 324.4 40.55
9 1 9 321.5 35.72
10 1 10 350.4 35.04
20 1 20 671.3 33.57
30 1 30 1199.7 39.99
40 1 40 1268.8 31.72
50 1 50 41008.9 820.18
60 1 60 40988.4 683.14
70 1 70 41105.4 587.22
80 1 80 81525.5 1019.07
90 1 90 81985.8 910.95
100 1 100 82014.2 820.14
200 1 200 42495.5 212.48
300 1 300 42576.4 141.92
400 1 400 42492.9 106.23
500 1 500 42988.4 85.98
600 1 600 43429.9 72.38
700 1 700 46641.0 66.63
800 1 800 53992.6 67.49
900 1 900 46058.4 51.18
1000 1 1000 47202.2 47.20
2000 1 2000 57735.4 28.87
3000 1 3000 67637.4 22.55
4000 1 4000 86611.8 21.65
5000 1 5000 96214.0 19.24
# Cumulative distribution of time for a single client to read a
# random 100-byte object from a single server. Each line indicates
# that a given fraction of all reads took at most a given time
# to complete.
# Generated by 'clusterperf.py readDist'
#
# Time (usec) Cum. Fraction
#---------------------------
0.00 0.000
58.18 0.000
63.41 0.010
73.98 0.020
74.33 0.030
74.55 0.040
74.69 0.050
74.88 0.060
75.13 0.070
75.59 0.080
77.86 0.090
81.09 0.100
88.03 0.110
92.61 0.120
93.05 0.130
93.47 0.140
94.85 0.150
95.38 0.160
95.63 0.170
95.81 0.180
95.98 0.190
96.15 0.200
96.37 0.210
96.72 0.220
97.79 0.230
98.38 0.240
98.95 0.250
100.49 0.260
105.88 0.270
114.86 0.280
122.11 0.290
125.41 0.300
125.78 0.310
125.89 0.320
125.96 0.330
126.01 0.340
126.09 0.350
126.27 0.360
126.65 0.370
130.34 0.380
132.00 0.390
133.17 0.400
133.41 0.410
134.16 0.420
134.57 0.430
134.64 0.440
134.72 0.450
134.83 0.460
134.88 0.470
134.91 0.480
134.93 0.490
134.95 0.500
134.96 0.510
134.98 0.520
134.99 0.530
135.00 0.540
135.02 0.550
135.03 0.560
135.04 0.570
135.05 0.580
135.06 0.590
135.07 0.600
135.09 0.610
135.10 0.620
135.11 0.630
135.12 0.640
135.14 0.650
135.16 0.660
135.18 0.670
135.20 0.680
135.23 0.690
135.27 0.700
135.32 0.710
135.36 0.720
135.39 0.730
135.42 0.740
135.45 0.750
135.48 0.760
135.51 0.770
135.54 0.780
135.58 0.790
135.63 0.800
135.71 0.810
135.80 0.820
135.88 0.830
136.00 0.840
136.28 0.850
137.08 0.860
139.10 0.870
139.50 0.880
139.66 0.890
139.92 0.900
140.20 0.910
140.95 0.920
141.64 0.930
142.88 0.940
147.48 0.950
160.34 0.960
171.64 0.970
181.10 0.980
185.58 0.990
203.16 0.999
278.85 0.9999
2521.43 1.000
# RAMCloud read throughput of a single server with a varying
# number of clients issuing individual reads on randomly
# chosen 100-byte objects with 30-byte keys
# Generated by 'clusterperf.py readThroughput'
#
# numClients throughput worker utiliz.
# (kreads/sec)
#-------------------------------------------
1 9 0.011
2 17 0.017
3 22 0.021
4 30 0.030
5 37 0.036
6 50 0.053
7 52 0.051
8 59 0.064
9 71 0.072
10 73 0.086
11 78 0.089
12 86 0.133
13 97 0.107
# RAMCloud read performance for 100 B objects
# with keys of various lengths.
# Generated by 'clusterperf.py readVaryingKeyLength'
#
# Key Length Latency (us) Bandwidth (MB/s)
#----------------------------------------------------------------------------
1 71.0 1.4
5 93.1 1.1
10 94.3 1.1
15 94.4 1.2
20 94.5 1.2
25 95.1 1.3
30 94.8 1.3
35 69.2 1.9
40 74.1 1.8
45 73.9 1.9
50 69.7 2.1
55 72.4 2.0
60 78.5 1.9
65 71.4 2.2
70 71.2 2.3
75 90.3 1.8
80 74.6 2.3
85 78.6 2.2
90 74.0 2.4
95 125.8 1.5
100 89.4 2.1
200 74.3 3.9
300 76.6 5.0
400 91.6 5.2
500 81.6 7.0
600 83.2 8.0
700 96.9 7.9
800 87.5 9.8
900 101.4 9.4
1000 95.0 11.0
2000 108.2 18.5
3000 141.4 20.9
4000 250.8 15.6
5000 277.2 17.5
6000 583.1 10.0
7000 584.1 11.6
8000 548.9 14.1
9000 500.7 17.3
10000 402.0 24.0
20000 666.9 28.7
30000 943.5 30.4
40000 760.4 50.3
50000 1006.3 47.5
60000 1217.8 47.1
# RAMCloud write performance for 100 B objects
# with keys of various lengths.
# Generated by 'clusterperf.py writeVaryingKeyLength'
#
# Key Length Latency (us) Bandwidth (MB/s)
#----------------------------------------------------------------------------
1 210.4 0.5
5 190.0 0.5
10 287.2 0.4
15 282.2 0.4
20 257.9 0.4
25 220.3 0.5
30 222.8 0.6
35 264.4 0.5
40 232.4 0.6
45 270.5 0.5
50 268.1 0.5
55 261.7 0.6
60 292.9 0.5
65 227.5 0.7
70 227.6 0.7
75 274.2 0.6
80 267.9 0.6
85 267.6 0.7
90 290.4 0.6
95 296.2 0.6
100 276.5 0.7
200 273.9 1.0
300 282.3 1.4
400 314.3 1.5
500 323.1 1.8
600 307.0 2.2
700 325.3 2.3
800 299.8 2.9
900 297.0 3.2
1000 313.5 3.3
2000 669.6 3.0
3000 1043.0 2.8
4000 1128.8 3.5
5000 811.7 6.0
6000 1225.7 4.7
7000 1212.3 5.6
8000 1434.1 5.4
9000 1609.4 5.4
10000 1669.2 5.8
20000 2475.1 7.7
30000 3149.6 9.1
40000 3660.7 10.4
50000 4213.2 11.3
60000 4945.5 11.6
basic.read100 87.3 us read single 100B object (30B key) median
basic.read100.min 50.0 us read single 100B object (30B key) minimum
basic.read100.9 99.8 us read single 100B object (30B key) 90%
basic.read100.99 144.7 us read single 100B object (30B key) 99%
basic.read100.999 180.8 us read single 100B object (30B key) 99.9%
basic.readBw100 1.2 MB/s bandwidth reading 100B object (30B key)
basic.read1K 97.2 us read single 1KB object (30B key) median
basic.read1K.min 80.4 us read single 1KB object (30B key) minimum
basic.read1K.9 140.4 us read single 1KB object (30B key) 90%
basic.read1K.99 150.1 us read single 1KB object (30B key) 99%
basic.read1K.999 188.4 us read single 1KB object (30B key) 99.9%
basic.readBw1K 9.4 MB/s bandwidth reading 1KB object (30B key)
basic.read10K 350.0 us read single 10KB object (30B key) median
basic.read10K.min 304.3 us read single 10KB object (30B key) minimum
basic.read10K.9 433.2 us read single 10KB object (30B key) 90%
basic.read10K.99 501.2 us read single 10KB object (30B key) 99%
basic.read10K.999 639.6 us read single 10KB object (30B key) 99.9%
basic.readBw10K 25.3 MB/s bandwidth reading 10KB object (30B key)
basic.read100K 1.4 ms read single 100KB object (30B key) median
basic.read100K.min 1.1 ms read single 100KB object (30B key) minimum
basic.read100K.9 1.4 ms read single 100KB object (30B key) 90%
basic.read100K.99 1.5 ms read single 100KB object (30B key) 99%
basic.read100K.999 1.6 ms read single 100KB object (30B key) 99.9%
basic.readBw100K 70.9 MB/s bandwidth reading 100KB object (30B key)
basic.read1M 8.6 ms read single 1MB object (30B key) median
basic.read1M.min 8.4 ms read single 1MB object (30B key) minimum
basic.read1M.9 8.7 ms read single 1MB object (30B key) 90%
basic.read1M.99 8.8 ms read single 1MB object (30B key) 99%
basic.readBw1M 111.2 MB/s bandwidth reading 1MB object (30B key)
basic.write100 219.4 us write single 100B object (30B key) median
basic.write100.min 151.1 us write single 100B object (30B key) minimum
basic.write100.9 290.2 us write single 100B object (30B key) 90%
basic.write100.99 321.6 us write single 100B object (30B key) 99%
basic.write100.999 1.7 ms write single 100B object (30B key) 99.9%
basic.writeBw100 413.3 KB/s bandwidth writing 100B object (30B key)
basic.write1K 308.7 us write single 1KB object (30B key) median
basic.write1K.min 220.6 us write single 1KB object (30B key) minimum
basic.write1K.9 340.5 us write single 1KB object (30B key) 90%
basic.write1K.99 404.0 us write single 1KB object (30B key) 99%
basic.write1K.999 1.6 ms write single 1KB object (30B key) 99.9%
basic.writeBw1K 3.1 MB/s bandwidth writing 1KB object (30B key)
basic.write10K 1.0 ms write single 10KB object (30B key) median
basic.write10K.min 861.5 us write single 10KB object (30B key) minimum
basic.write10K.9 1.2 ms write single 10KB object (30B key) 90%
basic.write10K.99 1.4 ms write single 10KB object (30B key) 99%
basic.write10K.999 3.8 ms write single 10KB object (30B key) 99.9%
basic.writeBw10K 8.7 MB/s bandwidth writing 10KB object (30B key)
basic.write100K 4.6 ms write single 100KB object (30B key) median
basic.write100K.min 4.4 ms write single 100KB object (30B key) minimum
basic.write100K.9 4.8 ms write single 100KB object (30B key) 90%
basic.write100K.99 5.6 ms write single 100KB object (30B key) 99%
basic.writeBw100K 20.6 MB/s bandwidth writing 100KB object (30B key)
basic.write1M 34.2 ms write single 1MB object (30B key) median
basic.write1M.min 34.0 ms write single 1MB object (30B key) minimum
basic.write1M.9 35.1 ms write single 1MB object (30B key) 90%
basic.write1M.99 35.4 ms write single 1MB object (30B key) 99%
basic.writeBw1M 27.8 MB/s bandwidth writing 1MB object (30B key)
# RAMCloud multiRead performance for 100 B objects with 30 byte keys
# located on a single master.
# Generated by 'clusterperf.py multiRead_oneMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 85.1 85.15
2 1 2 100.2 50.10
3 1 3 100.7 33.56
4 1 4 111.4 27.85
5 1 5 111.9 22.37
6 1 6 108.0 18.00
7 1 7 113.5 16.22
8 1 8 118.8 14.85
9 1 9 113.9 12.66
10 1 10 138.1 13.81
20 1 20 189.1 9.45
30 1 30 288.5 9.62
40 1 40 607.1 15.18
50 1 50 40607.9 812.16
60 1 60 40308.6 671.81
70 1 70 40503.8 578.63
80 1 80 40953.8 511.92
90 1 90 41058.0 456.20
100 1 100 41001.4 410.01
200 1 200 2036.6 10.18
300 1 300 1832.0 6.11
400 1 400 1741.8 4.35
500 1 500 1734.0 3.47
600 1 600 1918.2 3.20
700 1 700 1737.0 2.48
800 1 800 20073.2 25.09
900 1 900 2491.7 2.77
1000 1 1000 2898.7 2.90
2000 1 2000 14551.3 7.28
3000 1 3000 5874.4 1.96
4000 1 4000 10100.5 2.53
5000 1 5000 12620.2 2.52
# RAMCloud multiRead performance for 100 B objects with 30 byte keys
# with one object located on each master.
# Generated by 'clusterperf.py multiRead_oneObjectPerMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 101.3 101.28
2 2 1 114.6 57.28
3 3 1 145.2 48.39
4 4 1 156.5 39.13
5 5 1 155.4 31.07
6 6 1 153.2 25.53
7 7 1 152.6 21.80
8 8 1 170.9 21.36
9 9 1 188.2 20.91
10 10 1 164.8 16.48
11 11 1 164.0 14.91
12 12 1 207.2 17.27
13 13 1 233.0 17.92
14 14 1 298.9 21.35
15 15 1 295.3 19.69
16 16 1 288.6 18.04
17 17 1 281.4 16.55
18 18 1 320.1 17.78
# RAMCloud multi-read throughput of a single server with a
# varying number of clients issuing 70-object multi-reads on
# randomly-chosen 100-byte objects with 30-byte keys
# Generated by 'clusterperf.py multiReadThroughput'
#
# numClients throughput worker utiliz.
# (kreads/sec)
#-------------------------------------------
1 2 0.001
2 4 0.002
3 5 0.002
4 7 0.003
5 9 0.004
6 10 0.005
7 12 0.005
8 14 0.006
9 16 0.007
10 18 0.007
11 19 0.008
12 21 0.009
13 23 0.010
# RAMCloud multiWrite performance for 100 B objects with 30 byte keys
# located on a single master.
# Generated by 'clusterperf.py multiWrite_oneMaster'
#
# Num Objs Num Masters Objs/Master Latency (us) Latency/Obj (us)
#----------------------------------------------------------------------------
1 1 1 242.4 242.37
2 1 2 287.2 143.59
3 1 3 299.4 99.81
4 1 4 351.3 87.82
5 1 5 305.6 61.11
6 1 6 330.6 55.09
7 1 7 345.1 49.30
8 1 8 324.4 40.55
9 1 9 321.5 35.72
10 1 10 350.4 35.04
20 1 20 671.3 33.57
30 1 30 1199.7 39.99
40 1 40 1268.8 31.72
50 1 50 41008.9 820.18
60 1 60 40988.4 683.14
70 1 70 41105.4 587.22
80 1 80 81525.5 1019.07
90 1 90 81985.8 910.95
100 1 100 82014.2 820.14
200 1 200 42495.5 212.48
300 1 300 42576.4 141.92
400 1 400 42492.9 106.23
500 1 500 42988.4 85.98
600 1 600 43429.9 72.38
700 1 700 46641.0 66.63
800 1 800 53992.6 67.49
900 1 900 46058.4 51.18
1000 1 1000 47202.2 47.20
2000 1 2000 57735.4 28.87
3000 1 3000 67637.4 22.55
4000 1 4000 86611.8 21.65
5000 1 5000 96214.0 19.24
# Cumulative distribution of time for a single client to read a
# random 100-byte object from a single server. Each line indicates
# that a given fraction of all reads took at most a given time
# to complete.
# Generated by 'clusterperf.py readDist'
#
# Time (usec) Cum. Fraction
#---------------------------
0.00 0.000
58.18 0.000
63.41 0.010
73.98 0.020
74.33 0.030
74.55 0.040
74.69 0.050
74.88 0.060
75.13 0.070
75.59 0.080
77.86 0.090
81.09 0.100
88.03 0.110
92.61 0.120
93.05 0.130
93.47 0.140
94.85 0.150
95.38 0.160
95.63 0.170
95.81 0.180
95.98 0.190
96.15 0.200
96.37 0.210
96.72 0.220
97.79 0.230
98.38 0.240
98.95 0.250
100.49 0.260
105.88 0.270
114.86 0.280
122.11 0.290
125.41 0.300
125.78 0.310
125.89 0.320
125.96 0.330
126.01 0.340
126.09 0.350
126.27 0.360
126.65 0.370
130.34 0.380
132.00 0.390
133.17 0.400
133.41 0.410
134.16 0.420
134.57 0.430
134.64 0.440
134.72 0.450
134.83 0.460
134.88 0.470
134.91 0.480
134.93 0.490
134.95 0.500
134.96 0.510
134.98 0.520
134.99 0.530
135.00 0.540
135.02 0.550
135.03 0.560
135.04 0.570
135.05 0.580
135.06 0.590
135.07 0.600
135.09 0.610
135.10 0.620
135.11 0.630
135.12 0.640
135.14 0.650
135.16 0.660
135.18 0.670
135.20 0.680
135.23 0.690
135.27 0.700
135.32 0.710
135.36 0.720
135.39 0.730
135.42 0.740
135.45 0.750
135.48 0.760
135.51 0.770
135.54 0.780
135.58 0.790
135.63 0.800
135.71 0.810
135.80 0.820
135.88 0.830
136.00 0.840
136.28 0.850
137.08 0.860
139.10 0.870
139.50 0.880
139.66 0.890
139.92 0.900
140.20 0.910
140.95 0.920
141.64 0.930
142.88 0.940
147.48 0.950
160.34 0.960
171.64 0.970
181.10 0.980
185.58 0.990
203.16 0.999
278.85 0.9999
2521.43 1.000
# RAMCloud read throughput of a single server with a varying
# number of clients issuing individual reads on randomly
# chosen 100-byte objects with 30-byte keys
# Generated by 'clusterperf.py readThroughput'
#
# numClients throughput worker utiliz.
# (kreads/sec)
#-------------------------------------------
1 9 0.011
2 17 0.017
3 22 0.021
4 30 0.030
5 37 0.036
6 50 0.053
7 52 0.051
8 59 0.064
9 71 0.072
10 73 0.086
11 78 0.089
12 86 0.133
13 97 0.107
# RAMCloud read performance for 100 B objects
# with keys of various lengths.
# Generated by 'clusterperf.py readVaryingKeyLength'
#
# Key Length Latency (us) Bandwidth (MB/s)
#----------------------------------------------------------------------------
1 71.0 1.4
5 93.1 1.1
10 94.3 1.1
15 94.4 1.2
20 94.5 1.2
25 95.1 1.3
30 94.8 1.3
35 69.2 1.9
40 74.1 1.8
45 73.9 1.9
50 69.7 2.1
55 72.4 2.0
60 78.5 1.9
65 71.4 2.2
70 71.2 2.3
75 90.3 1.8
80 74.6 2.3
85 78.6 2.2
90 74.0 2.4
95 125.8 1.5
100 89.4 2.1
200 74.3 3.9
300 76.6 5.0
400 91.6 5.2
500 81.6 7.0
600 83.2 8.0
700 96.9 7.9
800 87.5 9.8
900 101.4 9.4
1000 95.0 11.0
2000 108.2 18.5
3000 141.4 20.9
4000 250.8 15.6
5000 277.2 17.5
6000 583.1 10.0
7000 584.1 11.6
8000 548.9 14.1
9000 500.7 17.3
10000 402.0 24.0
20000 666.9 28.7
30000 943.5 30.4
40000 760.4 50.3
50000 1006.3 47.5
60000 1217.8 47.1
# RAMCloud write performance for 100 B objects
# with keys of various lengths.
# Generated by 'clusterperf.py writeVaryingKeyLength'
#
# Key Length Latency (us) Bandwidth (MB/s)
#----------------------------------------------------------------------------
1 210.4 0.5
5 190.0 0.5
10 287.2 0.4
15 282.2 0.4
20 257.9 0.4
25 220.3 0.5
30 222.8 0.6
35 264.4 0.5
40 232.4 0.6
45 270.5 0.5
50 268.1 0.5
55 261.7 0.6
60 292.9 0.5
65 227.5 0.7
70 227.6 0.7
75 274.2 0.6
80 267.9 0.6
85 267.6 0.7
90 290.4 0.6
95 296.2 0.6
100 276.5 0.7
200 273.9 1.0
300 282.3 1.4
400 314.3 1.5
500 323.1 1.8
600 307.0 2.2
700 325.3 2.3
800 299.8 2.9
900 297.0 3.2
1000 313.5 3.3
2000 669.6 3.0
3000 1043.0 2.8
4000 1128.8 3.5
5000 811.7 6.0
6000 1225.7 4.7
7000 1212.3 5.6
8000 1434.1 5.4
9000 1609.4 5.4
10000 1669.2 5.8
20000 2475.1 7.7
30000 3149.6 9.1
40000 3660.7 10.4
50000 4213.2 11.3
60000 4945.5 11.6