The definition of availability is a topic often debated when organizations are defining metrics for their systems to be measured against. A survey by Continuity Central stated that 40 percent of IT professionals do not understand what qualifies as high availability. Surprisingly, the most important aspect of availability monitoring is frequently overlooked: Can the users access and utilize the application?
I’ve seen countless line-of-business owners call a meeting with the administrators of each component within a business application to make long lists of metrics that will define availability. Each team will throw a few items in the bucket, and advanced IT organizations will even bring all these items into one central console to look at everything at the same time. The list of what the administrators pull together often includes some of the following:
- Network – Link Saturation, Router Interface Status, Loss Rates
- Server – CPU, Memory Utilization, Context Switches, Queue Length
- Database – Table Space, Locks, Connection Time, Hit Ratio
- Web server – Process Up/Down, Log Events, Session Counts, Pings
One thing that you’ll notice above, is the lack of anything related to usability — a “so what?” metric.
If you are a SaaS provider, can you answer the question “Can my users log in?”; or if you are an eCommerce company “Can my customers complete a purchase?” To answer these types of questions you will need to watch users complete these processes or have test with automated synthetic users going through the motions.
If you’re going to spend the time and money to track application availability, make sure you understand if users can access the application or not.