How many of us IT professionals have been in a meeting similar to this: The chairs of various departments throughout the company are sitting around a long table and are giving a monthly summary. IT presents that the applications, network and servers were some amount of 9’s available and may explain an outage. The meeting goes on and then one of the heads explains a failure to meet department goals by stating some application was "slow."
IT is asked about it but unfortunately can only present data upon number of tickets and general up time. The slow comment is then picked up another department and IT is left in the untenable position of defending its metrics and supposedly achieved goals while other departments are blaming IT for lack of productivity.
The real problem is one of communication of expectations. IT has data that supports availability but the customer is complaining of slowness. Slowness is a subjective term and for IT to resolve the difficulty different metrics and SLAs are needed. Fortunately, there is a perfectly good way to measure slowness – time. When we think of availability we need to understand we are actually speaking of capacity while the users are interested in throughput.
By measuring transaction time (the amount of time it takes for the user to commit an action and receive the corresponding data from the program they are using) IT can state how fast an application is working in objective terms. SLAs can be established that some percentage of the transactions during a reporting period will be completed within a certain amount of time. This allows business decisions based upon performance and is a salve for the mysterious "slow" comment.
Availability is one of the early metrics IT has used to create a simple number to represent complex systems.
Robin Lyon is Director of Analytics at AppEnsure.