When you are working in IT Operations, your customer is the person who sends a request. Think back to being in front of a bank cashier or at a checkout counter when you insert you card. What do you expect? You expect:
■ the correct response (I took your money)
■ within the expected time, usually a couple of seconds.
Those are the two main benefits your data center provides. If you don't provide those benefits your business loses customers. It will filter down to you through sales, then marketing, then the CIO then – what are you going to do now?
If your data center is carefully monitoring all the hardware (servers, drives, routers, etc.) and all of the software (each OS, each process, each FIFO, etc.) you cannot detect if your customer is getting a response in the time they expect. You will see availability of the resources but not the quality of service or end-user experience.
You Have to Look at What the Customer Sees
You have to watch each request from your customer and measure the response time of the reply to that packet to see what your customer sees. That is the second benefit you provide to your customer.
This works only if you measure the response time of EVERY customer request and raise an intelligent, actionable alert when the response time deviates from what has been working just fine. When you collect this unique metric mentioned above, you can:
■ immediately inform affected customers that you know about the delay
■ immediately start working around and/or fixing the delay where it is occurring.
How do you immediately get the information you need to fix the problem? Go back to where I said, "an intelligent, actionable alert". The actionable alert should provide information regarding:
■ most likely cause of the problem
■ where the problem is
■ suggested fixes
Now within a few seconds you know what is needed to work around the problem to keep your customers happy and you know what to investigate for a long term solution.
Brad Denniston is a Senior Solutions Architect at AppEnsure.