Bêta de la nouvelle infrastructure VPS classic OVH

Bonsoir à tous,

Aujourd’hui j’ai enfin reçu les accès pour la bêta VPS d’OVH.

La nouvelle infrastructure VPS classic d’OVH est censée se trouvée une infrastructure semblable à celle du service Run Above.

Pour tester les vps je me suis servi de BYTE UNIX Benchmarks en version 5.1.3 le tout sur 2 Debian 7 à jour.

Je compare un VPS classic 1 sur l’infrastructure actuelle avec le vps classic 1 qui m’a été mis à disposition par OVH.

Test des performances CPU :

VPS Classic 1 de génération actuelle :

========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)

System: vps88237: GNU/Linux
OS: GNU/Linux — 2.6.32-042stab092.3 — #1 SMP Sun Jul 20 13:27:24 MSK 2014
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap= »UTF-8″, collate= »UTF-8″)
CPU 0: AMD Opteron(tm) Processor 4386 (6200.0 bogomips)
Hyper-Threading, x86-64, MMX, AMD MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
22:01:27 up 4 min, 1 user, load average: 0.09, 0.04, 0.01; runlevel 2

————————————————————————
Benchmark Run: mer. sept. 10 2014 22:01:27 – 22:29:34
1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 27363757.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3488.3 MWIPS (10.0 s, 7 samples)
Execl Throughput 3907.7 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 473017.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 157834.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 990727.8 KBps (30.0 s, 2 samples)
Pipe Throughput 1240011.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 192977.6 lps (10.0 s, 7 samples)
Process Creation 8084.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 6131.3 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 862.5 lpm (60.0 s, 2 samples)
System Call Overhead 1361994.1 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 27363757.3 2344.8
Double-Precision Whetstone 55.0 3488.3 634.2
Execl Throughput 43.0 3907.7 908.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 473017.1 1194.5
File Copy 256 bufsize 500 maxblocks 1655.0 157834.9 953.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 990727.8 1708.2
Pipe Throughput 12440.0 1240011.4 996.8
Pipe-based Context Switching 4000.0 192977.6 482.4
Process Creation 126.0 8084.7 641.6
Shell Scripts (1 concurrent) 42.4 6131.3 1446.1
Shell Scripts (8 concurrent) 6.0 862.5 1437.5
System Call Overhead 15000.0 1361994.1 908.0
========
System Benchmarks Index Score 1036.2

Résultats Ancien VPS

VPS classic 1 de la bêta :

========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)

System: vps: GNU/Linux
OS: GNU/Linux — 2.6.32-042stab092.3 — #1 SMP Sun Jul 20 13:27:24 MSK 2014
Machine: x86_64 (unknown)
Language: en_US.utf8 (charmap= »UTF-8″, collate= »UTF-8″)
CPU 0: Intel Xeon E312xx (Sandy Bridge) (6000.0 bogomips)
x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, I ntel virtualization
21:46:10 up 2 days, 6:19, 1 user, load average: 0.44, 0.19, 0.07; runlevel 2

————————————————————————
Benchmark Run: Wed Sep 10 2014 21:46:11 – 22:14:26
1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 37621247.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4483.9 MWIPS (9.9 s, 7 samples)
Execl Throughput 4813.3 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 804614.7 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 225679.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 2463999.1 KBps (30.0 s, 2 samples)
Pipe Throughput 1710742.3 lps (10.0 s, 7 samples)
Pipe-based Context Switching 268833.6 lps (10.0 s, 7 samples)
Process Creation 13181.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 8341.5 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1089.7 lpm (60.0 s, 2 samples)
System Call Overhead 2545185.5 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 37621247.6 3223.8
Double-Precision Whetstone 55.0 4483.9 815.3
Execl Throughput 43.0 4813.3 1119.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 804614.7 2031.9
File Copy 256 bufsize 500 maxblocks 1655.0 225679.8 1363.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 2463999.1 4248.3
Pipe Throughput 12440.0 1710742.3 1375.2
Pipe-based Context Switching 4000.0 268833.6 672.1
Process Creation 126.0 13181.8 1046.2
Shell Scripts (1 concurrent) 42.4 8341.5 1967.3
Shell Scripts (8 concurrent) 6.0 1089.7 1816.2
System Call Overhead 15000.0 2545185.5 1696.8
========
System Benchmarks Index Score 1557.0

Résultats nouveaux VPS

Résultats des performances CPU :

On gagne 60% de puissance au niveau du processeur, c’est très appréciable, on verra ce que ça donne à la fin de la bêta en production 😉

Ajout du 11/09/2014 – Test des performances réseaux :

Installation du test :

cd /tmp
wget -O speedtest-cli https://raw.github.com/sivel/speedtest-cli/master/speedtest_cli.py
chmod +x speedtest-cli
./speedtest-cli

Les 2 test sont fait sur le vps de la bêta.

Test 1 :

Retrieving speedtest.net configuration...
Retrieving speedtest.net server list...
Testing from OVH SAS (92.222.35.83)...
Selecting best server based on latency...
Hosted by Alionis (Paris) [1.30 km]: 6.404 ms
Testing download speed........................................
Download: 128.23 Mbits/s
Testing upload speed..................................................
Upload: 106.45 Mbits/s

Test 2 :

Retrieving speedtest.net configuration...
Retrieving speedtest.net server list...
Testing from OVH SAS (92.222.35.83)...
Selecting best server based on latency...
Hosted by NEOTELECOMS (Paris) [1.30 km]: 4.98 ms
Testing download speed........................................
Download: 129.69 Mbits/s
Testing upload speed..................................................
Upload: 117.08 Mbits/s

Conclusion sur les performances réseaux :

Les performances sont semblables à celles que j’ai sur l’infrastructure VPS actuelle.