As an IT professional you all probably have been in this situation, you found a software or a service you think would be great to your company and you instantly see the value (if it’s an open source than even better because there is no licensing cost attached), so you spin up a VM, install everything and voilà, you have a nice service up and running, now you have to convince everyone to start using it …
So, you send out an email, talk about it in the company meeting … and that’s mostly it …
How do you know if anyone is actually using it? when are they using it? is it up / down? Is it working in an efficient way? how can you extract any insights on how to make it even better? and let’s not forget, how do you present it to your management and get recognition???
if you don’t monitor and measure, you basically have nothing other than people saying “yeah, it’s nice, I use it once in a while ….”
So, let’s go back to the start, planning: you have an idea of a new service to introduce, these are the first 3 things you should think about:
- how do I monitor it?
- how do I react to any monitoring issues?
- how do I measure my service?
Step 1, monitoring: Monitoring is an obvious thing today, the market is full with monitoring software and services that can monitor almost everything you need, so sit down and have a list, sum up all the all the services you want to monitor, start in the OS level and climb your way up along to the application level, make sure you cover everything !
Step 2, automate: you probably have dozens and dozens of services running in your environment, so the last thing you want to do is look at monitoring dashboards all day or
Get tons of alerting emails you need to look at. You want to make your alerts efficient, so there is a dam good reason why you get a pager duty alert at 2 AM….
This is the reason why you would want to put automation when possible. you monitor your disk space? make sure that you have an automation in place to clean it up when reaching a certain threshold, monitoring your service state? have a watch dog to restart it if it crashes etc.
Obviously we can’t automate everything, but let’s do our best to automate whatever we can, you’ll thank me later …..
Step 3 – measure – now this is the most important thing of all, measurements will give you so much insights on how your application is used and how it can do better, you would never understand how did you do without it, so start by creating a base line, something that you can compare against, measurements are relative so you must have something you can compare against. start capturing information from whatever sources you can, logs, monitoring systems, employee surveys and have a baseline ready, now sit back and relax, wait a month or two and gather that information again, compare it to your baseline and analyze it, you would be amazed of the amount of insights you can extract and act upon, just as an example :
How many people have been using my service? (maybe I need to do a better PR for it, or maybe this service isn’t really needed as I thought it would be …)
What is the average load time? (maybe a faster CPU is needed? perhaps I need to do a better job optimizing the system).
After you have all the analytics ready, present it to your management, explain how did the business benefit from your service, and have everything you present backed up with numbers, no one can argue with numbers!
Make sure you make measurements a routine, as the service owner it’s your responsibility, so make sure your on top of things, making sure your service is used and brings value, and if not, perhaps it’s time to sunset it, we don’t want to end up with a lot of useless services that we need to maintain and don’t bring any value.
To sum things up, I know it’s easier to spin services up and forget about them, but in the long run it’s an awful mistake, consider every new service you introduce, plan it, monitor it, put automation around it and measure it, this way you would end up with only significant services that actually bring value to your business, and as a bonus, you’ll get recognition along the way .