いろんな記事でみんなやってるけど自分でもやってみる。
- 基本的にチューニングなし
- 東京リージョン
- 無料枠
- コンソールでUbuntu16.04起動してそのままベンチとる(なにもさわらない)
- 3回くらい打ってみて、誤差が10%くらいだったら1番いい結果を採用
- Unixベンチ(CPU) と インターネットへのアクセス速度、同セグメント同士の内部通信速度
- 家にちょうどCPU 2CoreのLinuxサーバ(1万6千円)いたのでついでに計測
とりあえず結果
| 環境 | マシンタイプ | 仮想 CPU 数 | メモリ | 料金（1時間） | １Core/1ヶ月 | UnixBench | 回線速度(ダウンロード) | 回線速度(アップロード) | 内部通信 |
|---|---|---|---|---|---|---|---|---|---|
| GCP | n1-standard-2 | 2 | 7.5GB | $0.13 | $47.74 | 2412.7 | 513.41Mbps | 268.80 Mbps | 3.91 Gbps |
| EC2 | m4.large | 2 | 8GB | $0.14 | $50.04 | 2170.1 | 251.58 Mbps | 173.81 Mbps | 568 Mbps |
| 自宅 | PRIMERGY TX100 S3P | 2 | 4GB | $0 | $6 12分割払い 電気代考えない |
3321.5 | 79.83Mbps | 35.83 Mbps | - |
コスト: 安鯖 >> GCE > EC2
CPU性能: 安鯖 >> GCE > EC2
外部通信速度: GCE > EC2 >> OCNマンションプラン
内部通信速度: GCE > EC2
想像以上にNTT-Xで衝動買いしたTX100が早かった。。。
Unixベンチ
Ubuntu16.04で こちら のツールの ./Run を実行
GCE
Benchmark Run: Mon Dec 26 2016 04:38:32 - 05:06:36
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 32889913.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4158.4 MWIPS (9.9 s, 7 samples)
Execl Throughput 3813.4 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1163952.7 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 309796.7 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3187773.4 KBps (30.0 s, 2 samples)
Pipe Throughput 2382718.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 99231.4 lps (10.0 s, 7 samples)
Process Creation 10327.3 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 9612.0 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1580.4 lpm (60.0 s, 2 samples)
System Call Overhead 3780852.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 32889913.9 2818.3
Double-Precision Whetstone 55.0 4158.4 756.1
Execl Throughput 43.0 3813.4 886.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 1163952.7 2939.3
File Copy 256 bufsize 500 maxblocks 1655.0 309796.7 1871.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 3187773.4 5496.2
Pipe Throughput 12440.0 2382718.4 1915.4
Pipe-based Context Switching 4000.0 99231.4 248.1
Process Creation 126.0 10327.3 819.6
Shell Scripts (1 concurrent) 42.4 9612.0 2267.0
Shell Scripts (8 concurrent) 6.0 1580.4 2634.0
System Call Overhead 15000.0 3780852.8 2520.6
========
System Benchmarks Index Score 1623.4
------------------------------------------------------------------------
Benchmark Run: Mon Dec 26 2016 05:06:36 - 05:34:42
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 41560354.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 7322.8 MWIPS (10.0 s, 7 samples)
Execl Throughput 6853.3 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1271128.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 335093.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3914204.4 KBps (30.0 s, 2 samples)
Pipe Throughput 2586059.3 lps (10.0 s, 7 samples)
Pipe-based Context Switching 446484.4 lps (10.0 s, 7 samples)
Process Creation 21331.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 12752.2 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1649.5 lpm (60.0 s, 2 samples)
System Call Overhead 5420983.2 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 41560354.9 3561.3
Double-Precision Whetstone 55.0 7322.8 1331.4
Execl Throughput 43.0 6853.3 1593.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 1271128.5 3209.9
File Copy 256 bufsize 500 maxblocks 1655.0 335093.4 2024.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 3914204.4 6748.6
Pipe Throughput 12440.0 2586059.3 2078.8
Pipe-based Context Switching 4000.0 446484.4 1116.2
Process Creation 126.0 21331.9 1693.0
Shell Scripts (1 concurrent) 42.4 12752.2 3007.6
Shell Scripts (8 concurrent) 6.0 1649.5 2749.2
System Call Overhead 15000.0 5420983.2 3614.0
========
System Benchmarks Index Score 2412.7
EC2
Benchmark Run: Mon Dec 26 2016 15:13:30 - 15:41:54
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 31797472.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3400.5 MWIPS (9.9 s, 7 samples)
Execl Throughput 4030.8 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1101694.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 295454.1 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3092616.0 KBps (30.0 s, 2 samples)
Pipe Throughput 2132125.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 45794.5 lps (10.0 s, 7 samples)
Process Creation 9804.2 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 9879.6 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1540.6 lpm (60.0 s, 2 samples)
System Call Overhead 3325235.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 31797472.0 2724.7
Double-Precision Whetstone 55.0 3400.5 618.3
Execl Throughput 43.0 4030.8 937.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 1101694.9 2782.1
File Copy 256 bufsize 500 maxblocks 1655.0 295454.1 1785.2
File Copy 4096 bufsize 8000 maxblocks 5800.0 3092616.0 5332.1
Pipe Throughput 12440.0 2132125.2 1713.9
Pipe-based Context Switching 4000.0 45794.5 114.5
Process Creation 126.0 9804.2 778.1
Shell Scripts (1 concurrent) 42.4 9879.6 2330.1
Shell Scripts (8 concurrent) 6.0 1540.6 2567.7
System Call Overhead 15000.0 3325235.7 2216.8
========
System Benchmarks Index Score 1447.6
------------------------------------------------------------------------
Benchmark Run: Mon Dec 26 2016 15:41:54 - 16:10:22
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 39501029.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 6052.8 MWIPS (10.1 s, 7 samples)
Execl Throughput 6302.2 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1195681.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 314091.6 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3607481.0 KBps (30.0 s, 2 samples)
Pipe Throughput 2347094.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 335122.9 lps (10.0 s, 7 samples)
Process Creation 18507.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 12197.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1584.2 lpm (60.0 s, 2 samples)
System Call Overhead 4767938.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 39501029.1 3384.8
Double-Precision Whetstone 55.0 6052.8 1100.5
Execl Throughput 43.0 6302.2 1465.6
File Copy 1024 bufsize 2000 maxblocks 3960.0 1195681.5 3019.4
File Copy 256 bufsize 500 maxblocks 1655.0 314091.6 1897.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 3607481.0 6219.8
Pipe Throughput 12440.0 2347094.6 1886.7
Pipe-based Context Switching 4000.0 335122.9 837.8
Process Creation 126.0 18507.5 1468.9
Shell Scripts (1 concurrent) 42.4 12197.8 2876.9
Shell Scripts (8 concurrent) 6.0 1584.2 2640.4
System Call Overhead 15000.0 4767938.8 3178.6
========
System Benchmarks Index Score 2170.1
ネットワークベンチ
SpeedTestを利用してベンチを取ってみる
ちなみにどっちも東京Regionから
http://archive.getdeb.net/getdeb/ubuntu/pool/apps/s/speedtest-cli/python-speedtest-cli_1.0.1-1~getdeb1_all.deb
をダウンロードして speedtest-cli --server 6508 #Yokohama を打つだけ
念のため3回打っても同じくらいだった
AWSはちゃんと東京っぽいけどGCEから打つと at2wn (Yokohama) [8338.92 km] ？？？
特殊なネットワークなのかな？
GCE
root@instance-3:~# speedtest-cli --server 6508
Retrieving speedtest.net configuration...
Testing from Google Cloud Platform (104.198.127.68)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by at2wn (Yokohama) [8338.92 km]: 28.419 ms
Testing download speed................................................................................
Download: 513.41 Mbit/s
Testing upload speed....................................................................................................
Upload: 268.80 Mbit/s
EC2
$ speedtest-cli --server 6508
Retrieving speedtest.net configuration...
Testing from Amazon (52.197.14.75)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by at2wn (Yokohama) [24.99 km]: 36.746 ms
Testing download speed................................................................................
Download: 251.58 Mbit/s
Testing upload speed....................................................................................................
Upload: 173.81 Mbit/s
内部通信
iperfを使って内部ネットワーク(同セグ)でのベンチマークをとってみる
apt install iperf して iperf -s しているサーバに iperf -c 10.140.0.3 でアクセスしてみる
GCE
------------------------------------------------------------
Client connecting to 10.140.0.3, TCP port 5001
TCP window size: 45.0 KByte (default)
------------------------------------------------------------
[ 3] local 10.140.0.2 port 59318 connected with 10.140.0.3 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 4.55 GBytes 3.91 Gbits/sec
EC2
$ iperf -c 10.0.0.206
------------------------------------------------------------
Client connecting to 10.0.0.206, TCP port 5001
TCP window size: 325 KByte (default)
------------------------------------------------------------
[ 3] local 10.0.0.65 port 54782 connected with 10.0.0.206 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 678 MBytes 568 Mbits/sec
おまけ
1万6千円で買ったPRIMERGY TX100 S3P のベンチ結果(回線はフレッツ光＋OCN)
- CPU 2 Core
- メモリ 4GB
- HDD 10TB(ただのファイル置き場)
Unixベンチ
Benchmark Run: 月 12月 26 2016 17:44:45 - 18:12:57
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 36316846.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4558.6 MWIPS (9.9 s, 7 samples)
Execl Throughput 5422.4 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1139181.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 334280.7 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 2469074.4 KBps (30.0 s, 2 samples)
Pipe Throughput 2476910.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 233752.1 lps (10.0 s, 7 samples)
Process Creation 16076.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 12390.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2217.9 lpm (60.0 s, 2 samples)
System Call Overhead 4104422.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 36316846.0 3112.0
Double-Precision Whetstone 55.0 4558.6 828.8
Execl Throughput 43.0 5422.4 1261.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 1139181.1 2876.7
File Copy 256 bufsize 500 maxblocks 1655.0 334280.7 2019.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 2469074.4 4257.0
Pipe Throughput 12440.0 2476910.0 1991.1
Pipe-based Context Switching 4000.0 233752.1 584.4
Process Creation 126.0 16076.8 1275.9
Shell Scripts (1 concurrent) 42.4 12390.7 2922.3
Shell Scripts (8 concurrent) 6.0 2217.9 3696.5
System Call Overhead 15000.0 4104422.6 2736.3
========
System Benchmarks Index Score 1975.4
------------------------------------------------------------------------
Benchmark Run: 月 12月 26 2016 18:12:57 - 18:41:09
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 71781181.4 lps (10.0 s, 7 samples)
Double-Precision Whetstone 9004.1 MWIPS (9.9 s, 7 samples)
Execl Throughput 11455.9 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1368120.2 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 380983.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 3480067.5 KBps (30.0 s, 2 samples)
Pipe Throughput 4889444.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 770362.3 lps (10.0 s, 7 samples)
Process Creation 30494.1 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 18946.3 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2379.6 lpm (60.0 s, 2 samples)
System Call Overhead 6507685.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 71781181.4 6150.9
Double-Precision Whetstone 55.0 9004.1 1637.1
Execl Throughput 43.0 11455.9 2664.2
File Copy 1024 bufsize 2000 maxblocks 3960.0 1368120.2 3454.8
File Copy 256 bufsize 500 maxblocks 1655.0 380983.4 2302.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 3480067.5 6000.1
Pipe Throughput 12440.0 4889444.2 3930.4
Pipe-based Context Switching 4000.0 770362.3 1925.9
Process Creation 126.0 30494.1 2420.2
Shell Scripts (1 concurrent) 42.4 18946.3 4468.5
Shell Scripts (8 concurrent) 6.0 2379.6 3966.1
System Call Overhead 15000.0 6507685.5 4338.5
========
System Benchmarks Index Score 3321.5
speedtest
# speedtest-cli --server 6508
Retrieving speedtest.net configuration...
Testing from OCN (153.205.138.69)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by at2wn (Yokohama) [24.99 km]: 22.115 ms
Testing download speed................................................................................
Download: 79.83 Mbit/s
Testing upload speed...................................................................................
.................
Upload: 35.83 Mbit/s