(Or; Good morning server. And how are we feeling today?) In a SaaS environment uptime is everything. Rule is simple. No reliability, no credibility, no business. So, how do we know that all the different servers with all the different roles are happy and functional? How can we see when more server power, RAM or disk space is needed. Monitoring. A few years ago I developed our own server monitor system. Nothing fancy really. It mainly checked the servers every ten minutes to measure the response time. It was a client/server architecture where we had to develop small programs, one for each server role that did the necessary tests and responded back to the monitor server. The small apps did things like connecting to the database server, initiated a TTS voice, ran a php program to ensure that the web server was responding correctly etc. If the response was not the one the monitor expected, or if the response took longer then the set threshold, it was set to notify someone that was available 24 hours day, meaning at that time, me. A notification email along with an SMS was sent to my cell phone, and I could just interrupt whatever I was doing, find myself a computer with Internet connection and solve the problem. However, in the long run, this was not an ideal situation. I dared to set myself as receiver on all monitor alerts thanks to the fact that our servers almost never failed. Most of them ran without any interruptions for years. When the amount of servers grew, I had to make a decision. Should I spend some valuable management time to further develop my self made monitor system or should I grow up and finally take a look on what the marked had to offer. I choose the later. First step was to employ someone that could take care of all this. The problem building companies from scratch is that it tends to be difficult to delegate… The new guy did a market research and found out that OpManager was a pretty good system. Nice UI, good looking graphs, great alert possibilities and all that. We ran it for a couple of trial periods and were just about to purchase it for the whole environment when we got this IBM blade centre. IBM offers their monitor system for free. IBM Director Not only could it monitor low level stuff, it could also be used to install and update software on the different servers remotely. Also, it could do something that OpManager couldn’t; monitor the Storage Area Network file cluster as well as the blade centre itself (not only the blades). However, due to some shortcomings in the graphical reports, we settled for the OpManager. Time will tell if we made the right choice. Never underestimate the power of nice graphs showing what you want to see; how the systems are doing over time.