OVH Community, your new community space.

RPS vs KIMSUFI vs 100M Comparison


unnamed_crewman
14-12-2009, 17:22
Hi,

I thought the benchmarking was a good idea so I have downloaded phpspeed 1.0 beta benchmarking tool.

It can be downloaded from here : www.phpspeed.com

All you do is reconfigure the config file for database access, upload to a site then install.

You can run it from the browser – I tried 5 times for each test to average it out.

I have tried it with the following configurations of OVH shared hosts :

(1) Shared hosting 1G account with its own database and using php5 (I don’t think php4 runs this benchmark). OVH Cluster 14. mySQL version 5.0.88-log and php 5.2.11

(2) 60GP shared hosting from OVH Cluster 10, but using private mySQL with default my.cnf 128mB option. mySQL version 5.1.31 and php 5.2.11

Here are the results:

6 tests are conducted with descriptions.

Synthetic PHP BenchMark :
This set of benchmarks is designed to test various functions within PHP. The tests are synthetic in nature and will test the speed of your PHP install, but not necessarily the 'real world' performance of your site.
Synthetic MySQL BenchMark :
This set of benchmarks is designed to test how quickly mysql is reading and writing to your database. The tests are synthetic in nature and will test the speed of your mysql install, but not necessarily the 'real world' performance of your site.
Synthetic Read/Write BenchMark :
This tests measures how fast your server reads and writes to files on the hard drive. This test is a good indicator of the effectiveness of hardware tweaks. This test is also affected by the speed of your CPU, and the amount of memory you have.
Real World PHP BenchMark :
This benchmark measures the time to load 4 PHP pages of various sizes. There are no database pulls in this benchmark, this test will measure the real world time for someone to load a PHP page on your server.
Real World PHP & MySQL BenchMark :
This benchmark measures the time to load 4 PHP pages of various sizes including MySQL calls in each page. This is a 'real world' simulation of a dynamic web page like most of those on the web today.
Server Benchmark :
This benchmark throws everything we've got at your server to find out how fast it really is!

My results are as follows (more is better):

Synthetic PHP BenchMark :
1G w/o private SQL = 3575
60GP with private SQL = 3318
Synthetic MySQL BenchMark :
1G w/o private SQL = 891
60GP with private SQL =700
Synthetic Read/Write BenchMark :
1G w/o private SQL = 745
60GP with private SQL = 746
Real World PHP BenchMark :
1G w/o private SQL = 1237
60GP with private SQL = 1525
Real World PHP & MySQL BenchMark :
1G w/o private SQL = 626
60GP with private SQL = 552
Server Benchmark :
1G w/o private SQL = 664
60GP with private SQL = 774

Only conclusion I have is that they are very similar, as you would expect as they are running on similar clustered cpus – privateSQL improvement seems to be minimal – could try other settings if there are specifics that anyone recommends - but on this basis not worth the money.

So if there is anyone with a RPS, Kimsufi or dedicated server who wants to add data – please do so. Would hope that they are in the thousands!

Better still if OVH want to produce a comprehensive benchmark comparison for their offerings that would be a brilliant tool for planning for the non experts.

Obviously if anyone has a better benchmark then we could try that instead!

IainK
04-12-2009, 16:53
Hmm that is interesting. It certainly would be easy enough to run benchmarking software across various OVH servers and plot those results in graphs.
I could provide the results for the Kimsufi Q1-T

unnamed_crewman
04-12-2009, 11:22
Hi,

I came across this link on the French Forum so I thought I would throw it out to you. Admittedly it is in French, but you get the gist and some of the results are interesting.

It is clearly out of date - has anyone else tried or is able to try something similar? It would certainly be enlightening for those of us pondering upgrades - and there are lots more of these now with the new set up fees.

Alistair

http://rps.brunop.be/dossiers/perfor...urls&artpage=6