App Reliability in the Age of SAAS

As somebody who enjoys blowing away his notebook computer to install a new operating system every six weeks or so, I have a special appreciation for the way that software as a service lets me leave my key applications and data, accessible and undisturbed, in the cloud.

At least, “accessible and undisturbed” describes the way that things are supposed to be with SAAS, when the chain of components from browser to operating system to client hardware to Internet connectivity to the black box of your SAAS provider’s systems remains intact and performing as intended.

I just wrapped up a pair of stories in which I’ve attempted to flesh out the issues involved in achieving SAAS reliability, a combination of uptime and acceptable performance that can be much tougher to ensure with services that you don’t directly control:

In my research on the topic, it seemed to me that the discussion of measuring and ensuring SAAS performance, at least when approached from a customer perspective, is still in a relatively immature state. Most of what I encountered dealt with the perspective of the SAAS builder or vendor.

My findings boiled down to three main elements:

  • defining your expectations at the beginning of your SAAS evaluation process;
  • discussing reliability and architecture in depth with your prospective SAAS provider; and
  • testing the services, both through user piloting and through use of monitoring and performance tools that do not require intimate access to the SAAS application or infrastructure (which, by definition, you won’t have with SAAS).

I would be interested to learn about how eWEEK’s readers approach the problem of ensuring reliability of services, and to fold that feedback into a future update of this story. Also, beyond reliability, what are your primary SAAS concerns?