I have been hearing some criticism on Azure reliability lately. So, I decided to check out if anything I hear has any base by trying to compare Azure VM’s against other IAAS’s and I was quite surprised instability of Azure when putting side-by-side with its competitors.
Even though I don’t use Google Compute Engine personally. I decided to compare it together with EC2 and Azure VM’s since it is among one of the most popular ones.
I think Azure is very nice platform, I use it daily (Office 365 and VisualStudio.com) but they’ve effectively priced me out every time I need to get a server up quickly. Then I always end up to EC2 (Which I like also) but I would like to run some servers on VM’s. Just to get first hand experience how is it to operate real stuff on their beautifully designed Azure Admin.
Disclaimer: I have not used in-depth research on the topic. This is just what I found out by looking numbers from CloudHarmony. If you have more in-depth knowledge on the topic. Please comment the post. I would love to hear it.
Here are the numbers from CloudHarmony
1 Year Global Uptime
You can see that average downtime has been 3.3 hours per region where as EC2 had 14 minutes. But if you are running IAAS on here it makes much more sense to look about the uptime of the regions (datacenter). As a sample I put up Europe here to see same numbers from 1 year on Europe data centers.
1 Year Europe Uptime
This is actually looking worse for Azure and Google where as very good for Amazon.
If however you are planning to run IAAS on some of these cloud services, do your own research and include the specific services your VM’s are running. I recommend also paying close attention to the monthly/quarterly uptime trend in addition to aggregate uptime numbers. I would also recommend building your application/infrastructure cloud agnostic so you are not married to them (yeah, easy to say…).