In July, the technology review website Phoronix published several benchmarks they ran of Ubuntu 12.04 running on Amazon EC2. The benchmarking software Phoronix used is open source and it’s easy to replicate their tests, so we ran them on Brightbox Cloud as a comparison.
Each benchmark test is listed with the results for the two server types. Whenever the score was better, the result is in green (better might be lower or higher, depending on the type of test).
In the majority of cases, the Brightbox server performs better than the EC2 server, in some cases significantly better.
A detailed breakdown of all the results, including the other EC2 server types, are available on openbenchmarking.org.
I’ve included the graphs for some of the more notable results below.
34% more HTTP requests per second.
The Apache benchmarks show us leading with 34% more HTTP requests per second. Definitely a good choice for web deployments.
Up to 30% more GraphicsMagick operations per second.
If your web apps are doing image transformations, you’ll be pleased to see up to 30% more GraphicsMagick operations per second.
Up to 220% faster with NASA supercomputing benchmarks.
Those of you doing heavy computation win out the most, with the NASA supercomputing benchmarks coming in at over 220% faster at peak.
27% faster at compiling a Linux kernel.
We’re 27% faster at compiling a Linux kernel, which in itself probably won’t excite anyone (other than perhaps Gentoo users) but it’s a good overall test of cpu, memory and disk io.
Our “large” server type comes with 1GB more RAM and is about £43/month cheaper too. It does have less disk capacity, but our storage is hardware RAID6 with 15k rpm SAS disks and battery-backed write cache, giving high performance and high availability.
And as with all Brightbox Cloud servers, you’re not limited to any particular operating systems. We provide full virtualisation so you can pretty much run any x86/amd64 OS you want.
Of course, a single set of benchmarks is never enough to draw complete conclusions, we recommend you test for your own use cases and do you own comparisons (and do let us know how you get on!)
posted 20 Sep 2012 by John Leach