Benchmarking Cloud Servers – Just How Fast Are Our Servers?
Earlier this week we launched our highly anticipated cloud servers. Our team of engineers put countless hours in this new SherWeb offering, which they built with one focus in mind—performance.
Using the components of well-established brands like Intel and Dell, we developed what we hope will become the go-to solution for businesses seeking high performance cloud servers.
It’s one thing to claim our cloud servers’ performance can’t be matched by the big guys, but words alone can’t support such a bold statement. And we know you aren’t gullible enough to fall for a few well-written claims. So, how can we prove it?
We are so confident our cloud servers can beat the performance of other major providers that we ran all kinds of benchmarks to confirm it. Starting next Monday, and for a full week, we’ll be revealing one new benchmark a day. We’ll add the results to our website as well.
With each post, we’ll provide our analysis of the benchmark and why the data is important. Let’s get things started today with a key factor when benchmarking: Methodology! We want to be truly transparent as to how our tests were conducted.
Benchmarking against top providers
We’ve run our tests against well-known providers: Microsoft (Azure), Amazon (AWS), RackSpace and Digital Ocean.
These companies weren’t randomly chosen. We decided to compare ourselves to the providers that are getting the most attention in the cloud server industry. They offer compute-intensive or performance-oriented products, but can they live up to expectations? Are they truly as effective as touted?
To run the benchmarks, we provisioned servers on all of these providers’ platforms using what would be the most comparable specifications. Obviously, we picked the products these providers described as performance-focused or compute-centric.
We used Ubuntu 14.04 as the operating system for all the tests. Moreover, we made sure CPU and RAM were comparable and that we used SSD storage when available.
See the chart below for the list of specs for each provider.
|Azure||East – US||2||7 GB||X||Ubuntu 14.04||3.13.0-35-generic||D2 Standard||AMD Opteron 4171 HE @ 2.09GHz (2 cores)|
|AWS||US West (Oregon)||2||3.75 GB||X||Ubuntu 14.04||3.13.0-29-generic||C3.large||Intel Xeon E5-2680 v2 @ 2.80GHz (2 cores)|
|DigitalOcean||New York 3||2||4 GB||X||Ubuntu 14.04||3.13.0-24-generic||2 x Intel Xeon 35-2630L v2 @ 2.4GHz (2 cores)|
|RackSpace||Northern Virginia||2||2 GB||X||Ubuntu 14.04||3.13.0-30-generic||Performance 1||2 x Intel Xeon E5-2670 @ 2.6GHz (2 cores)|
|SherWeb||Montreal||2||4 GB||X||Ubuntu 14.04||3.13.0-24-generic||Intel Xeon E5-2690 v2 @ 3GHz (2 cores)|
Keeping that information in mind, you could easily replicate our tests and come to the same results we did. In fact, if you want even more details on how we provisioned each of these servers, all the configurations details are in this PDF-file.
To evaluate the performance of a cloud server, what should you test? We’re looking for speed in 3 key components:
For each component, you have to look for elements like latency, IOPS, throughput, bandwidth, etc. With results in hand; you can compare the scores of providers, according to their performance.
The critical element to remember is that every test is replicable. There’s no lying in benchmarking—it’s just like the experiments we used to do back during Science class in high school: we follow the step-by-step instructions; hence we should all achieve the same results.
What to expect from next week’s benchmarks
What are the key takeaways you’ll get from the benchmarks we’ll be sharing with you next week?
You’ll see that even when products are praised as “performance” cloud servers by the other providers, more often than not there’s no consistency in their results for speed or efficiency. Sometimes, even when they do well in one area, other providers beat them in the next test.
As mentioned, starting Monday we’ll be publishing one new benchmark every day next week. So stay tuned, our first post will feature benchmarks specifically relating to RAM.
We also invite you to sign up and test our cloud servers for yourself. It’s absolutely free and there are no contracts or obligations. You only pay for what you use.
Sell all of our benchmarks!
Day 1: Cloud Servers Benchmarks – Day 1 (RAMSpeed)
Day 2: Cloud Servers Benchmarks – Day 2 (Flexible IO Tester)
Day 3: Cloud Servers Benchmarks – Day 3 (C-Ray)
Day 4: Cloud Servers Benchmarks – Day 4 (FFmpeg)
Day 5: Cloud Servers Benchmarks – Day 5 (Apache HTTP Server)