The Inquirer-Home

The cloud is not smart enough to be reliable, says uptime specialist

Engineering, not marketing hype is required
Thu Aug 18 2011, 18:10

QUESTIONS need to be asked over whether the cloud is really up to the job of hosting critical services, according to resilience firm Neverfail.

Neverfail CTO Paddy Falls told The INQUIRER that following Amazon's recent outage firms need to ask "can you trust the cloud to give you sufficient resilience" in services? Amazon's most recent outage occurred last week, with initial reports suggesting that one of the firm's datacentres had succumbed to a lightning strike, only for that story to be changed to blame a dodgy electrical transformer.

Excuses aside, the downtime of Amazon's widely used cloud services left some web sites offline, something that many cloud users believe should not be possible thanks to the vast numbers of physical machines that make up a cloud. Falls suggests that companies shouldn't bank on the machines of a single provider to ensure resilience.

"Don't put all of your eggs in one cloud basket", said Falls, adding that customers should not "depend totally on just one provider". He said that in the foreseeable future most firms should look at having high availability services running locally with the cloud being used for disaster recovery.

Of particular interest was Falls' claim that today the weak link is not hardware but rather errors in configuration, adding that errors in configuring a virtual machine have a "ripple effect" on all services running on that virtual machine. While cloud providers, and to an extent hardware vendors, have been releasing ever more sophisticated hardware monitoring tools, Falls claims that what is needed is application level monitoring.

Falls also predicts system administrators will have to automate more and more tasks in order to keep up with the growing number of machines, both physical and virtual, in order to keep potential configuration errors to a minimum.

Even with better configuration, application level monitoring and general maturity in cloud deployment, Falls says that a self-healing cloud is still five years away. And for all the hype surrounding the cloud, the recent past has shown that companies clearly cannot rely on simply lobbing their applications onto a cloud and hoping that there is safety in numbers.

Falls' comments highlight the need for real software engineering when deploying services on the cloud rather than simply hoping that the marketing buzz surrounding the cloud is enough to keep their businesses ticking over. µ


Share this:

blog comments powered by Disqus
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

Existing User
Please fill in the field below to receive your profile link.
Sign-up for the INQBot weekly newsletter
Click here
INQ Poll

Microsoft Windows 10 poll

Which feature of Windows 10 are you most excited about?