What the heck is an m1.small anyway?

A lot of us in the cloud community enjoy reading what I refer to as "cloud trivia." Cloud trivia encompasses metrics of measured speed or throughput, potentially across cloud providers. A special favorite of mine is the Rackspace post [1] showing how much faster their 1.5c/hr server is than ec2's 8.5c/hr m1.small. There must be a way to use this data to make cloud offerings more relevant to application or budget requirements.

The idea of acting on performance measurements as opposed to cloud claims like vpu counts is hardly new. Last year, OCCI discussed and tabled the issue due to various reasons including the need for an independent source of metrics [2].

CloudHarmony could well be that independent source. Their blog proliferates performance comparisons across more clouds than I've ever seen on a single page [3]. Yesterday, I had a chat with CloudHarmony's Jason Read about what they are doing. The bottom line is that their goal is to be "the" source for independently measured performance results. While these results are in blog form today, they have ReST and SOAP interface development underway.

With a ReST interface to performance data, tools can correlate service offerings with test results to achieve the best bang for the buck. Moreover, this data establishes reasonable grounding for SLA definition of applications and platforms dependent on cloud resources. Numerous other developments will surface once rules engines access this data.

Watch this space, and expect more intelligent tooling for the cloud on the horizon.

[1] http://www.thebitsource.com/featured-posts/rackspace-cloud-servers-versu...
[2] http://www.ogf.org/pipermail/occi-wg/2009-October/001428.html
[3] http://blog.cloudharmony.com/