Microsoft Trails Amazon, Google In Cloud Reliability - InformationWeek

InformationWeek is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Cloud // Infrastructure as a Service
11:25 AM
Connect Directly

Microsoft Trails Amazon, Google In Cloud Reliability

CloudHarmony's benchmarking report on cloud service reliability shows Amazon shines in compute, Google in storage.

Amazon had the best uptime for cloud-based compute services in 2014, but Google had the best cloud storage uptime, according to CloudHarmony, a service founded five years ago to provide cloud benchmarking. Microsoft, the other big cloud supplier, trailed well behind, with some smaller suppliers showing better availability than Microsoft's Azure service did.

The report cards come from CloudHarmony's year-end statistics for infrastructure-as-a-service suppliers. Amazon's EC2 compute service sustained a 99.9974% uptime performance, translating to being down for a total 2.01 hours during the year.

Google's Compute Engine showed 99.9814% availability for the year, meaning it was out of commission for a total of 3.4 hours, according to the CloudHarmony report. Both offer customers service level agreements (SLAs) of 99.95%, so they appeared to stay out of the penalty box during the year. Amazon's SLA says it will provide 99.95% minimum level of availability each month, not averaged over the whole year, but it doesn't appear to have had any outage exceeding that monthly limit.

When it came to storage, the leadership was reversed. Google's Storage service was available 99.9996% of the time, with only 14.23 minutes of total downtime. Amazon's S3 storage, on the other hand, showed 99.9951% availability, with 2.69 hours of downtime.

Microsoft's Azure had a rougher year, with its compute service available 99.9374% of the time, with outages totaling 42.94 hours. Its Object Storage service was available 99.985% of the time, with downtime of 10.94 hours.

[Want to learn more about Microsoft's November outage? Read Microsoft Azure Outage Blamed On Bad Code.]

(Source: CloudHarmony)
(Source: CloudHarmony)

DigitalOcean, the fast-growing, New York-based service which boasts an SLA of 100% uptime, must have owed customers some time credits, since its uptime for the year came in at 99.9767%, with 9.64 hours of total downtime.

CenturyLink Cloud weighed in with an uptime of 99.9598%, for 27.06 hours of downtime total for the year.

Rackspace, which is still in the IaaS business but has pivoted toward managed services, posted a respectable record for its cloud infrastructure: 99.9892% uptime, with 40.36 minutes of downtime total.

Joyent kept pace with the market leaders with 99.9945% uptime and total downtime of 1.59 hours. Likewise BlueLock, a smaller, VMware-oriented supplier, kept pace with 99.9998% uptime and total outage of just 1.27 minutes.

GoGrid, which focused on high-performance compute service in 2014, showed 99.9944% uptime, with a total of 1.43 hours of downtime.

The HP Cloud posted one of the few perfect scores. CloudHarmony listed no statistics on HP's compute service, but its Object Storage was up 100% of the time with no downtime. 

Apply now for the 2015 InformationWeek Elite 100, which recognizes the most innovative users of technology to advance a company's business goals. Winners will be recognized at the InformationWeek Conference, April 27-28, 2015, at the Mandalay Bay in Las Vegas. Application period ends Jan. 16, 2015.

Charles Babcock is an editor-at-large for InformationWeek and author of Management Strategies for the Cloud Revolution, a McGraw-Hill book. He is the former editor-in-chief of Digital News, former software editor of Computerworld and former technology editor of Interactive ... View Full Bio

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Print  | 
More Insights
Newest First  |  Oldest First  |  Threaded View
David Wagner
David Wagner,
User Rank: Strategist
1/15/2015 | 5:31:19 PM
Re: Linode, CloudSigma, and SoftLayer
Right, I'm totally with you on 99% seeming good but not good enough. I just think year to year numbers may not be good enough. You could have an outage on Jan. 1st of one year, and go all the way to December 31 of the following year to your next one. That one day shy of two years without an outage but both year's numbers would look bad. We need some sort of longer look and also some sort of median time between outages as well. And also, which is worse-- 1 one hour outage or 60 one minute outages? 
Curt Franklin
Curt Franklin,
User Rank: Strategist
1/15/2015 | 4:15:47 PM
Re: Linode, CloudSigma, and SoftLayer
@Dave, reliability is one of those areas where precision matters a lot. 99% reliability means that you can have over 3.5 days of downtime in a year. That's a bunch. In general, the money (in terms of both cost to maintain and cost to contract for services) really starts building when you move between four and five nines. A whole day can make an enormous difference when it comes to the number of nines to the right of the decimal place.

Your question about variability is a good one: I'd love to know that myself!
David Wagner
David Wagner,
User Rank: Strategist
1/15/2015 | 2:06:59 PM
Re: Linode, CloudSigma, and SoftLayer
@Charlie- Obviously downtime is money and 40 hours versus 3 or 4 looks really big. It is ten times bigger, right? 

But I have just enough understanding of statistics to be dangerous. What is the variability of these numbers year to year? In other words, we know Amazon had a rough year two (three?) years ago when it had a multi-day outage. Can the same be said for any company? Jsut one bad day throwsyou from first to last in these rankings?

Because all of the companies here are at the 99% rate. A few hours seems to throw off the digits behind the decimal really fast.
Charlie Babcock
Charlie Babcock,
User Rank: Author
1/15/2015 | 1:42:50 PM
Linode, CloudSigma, and SoftLayer
The small New Jersey provider, Linode, posted good numbes with 99.9953% uptime and a total 1.23 hours downtime for the year. CloudSigma had 99.964% availability, with 5.1 hours of downtime. Not clear to me why stats weren't offered on IBM SoftLayer compute.
InformationWeek Is Getting an Upgrade!

Find out more about our plans to improve the look, functionality, and performance of the InformationWeek site in the coming months.

How CIO Roles Will Change: The Future of Work
Jessica Davis, Senior Editor, Enterprise Apps,  7/1/2021
A Strategy to Aid Underserved Communities and Fill Tech Jobs
Joao-Pierre S. Ruth, Senior Writer,  7/9/2021
10 Ways AI and ML Are Evolving
Lisa Morgan, Freelance Writer,  6/28/2021
White Papers
Register for InformationWeek Newsletters
Current Issue
Flash Poll