The cloud is all the rage today. For online business owners and startup entreprenurs, the cloud is often pitched as a low entry cost solution to many scalability challenges. Just throw your web application into the cloud and pay as you grow.

But does the cloud deliver? According to researchers at the University of New South Wales, the cloud may not be all that it’s cracked up to be. When put to stress tests, cloud computing solutions offered by Amazon, Google and Microsoft showed some weaknesses.

Stress tests that simulated 2,000 concurrent users connected to applications hosted on the Amazon EC2, Google AppEngine and Microsoft Azure cloud computing solutions did confirm that these “cloud services scale up and respond dynamically to…demand” but also revealed inconsistent performance. For instance, according to researcher Anna Liu, response times during the tests varied by a factor of 20 depending on the time of day the testing occurred. Not good.

The tests also revealed that the various solutions were uniquely suited to different types of applications. AppEngine, for example, appears best suited to simple applications, as “none of your data processing tasks can last any longer
than thirty seconds
“.

In all cases, Liu found that “none of the platforms have the kind of monitoring required to have a reasonable
conversation about performance
“. She noted that most of the tools offered by Amazon, Google and Microsoft were useful to developers but not to business users.

Obviously, reputable individuals have not claimed that cloud computing is an end-all and be-all solution. But the University of New South Wales research does add more weight to the argument that most of the cloud computing solutions out there have relatively narrow sweet spots and some significant limitations. For most online businesses and startups, simply throwing an application into the cloud isn’t necessarily a good, cost-effective strategy because in many cases, the type of dynamic scalability offered by cloud computing solutions is unnecessary and scalability without consistent performance is of dubious value.

Photo credit: Kables via Flickr.