There are a large number of tools dedicated to Service Management of IT infrastructure. Many of these lay claim to providing performance management or capacity planning to a business.
But how difficult a job is this, and is it worth spending money on such tools? Most operating systems contain bundled applications/tools that will report the current activity of a system. The purpose of a performance or capacity tool is to record this information on a regular basis, store it, and provide analysis tools for the specialist to use. There are limitations to the bundled applications/tools, and for this reason some vendors have written their own systems that can interrogate a running kernel directly. This will certainly provide additional information over and above the standard toolsets, but it is unnecessary if all you need to look at is overall CPU, Memory and Disk activity.
Tool vendors also promote their ability to report and analyse application data. Before you get excited about this, consider a very simple and fundamental fact. If your application developers have not included any instrumentation in their application, then where will the vendor’s tool get it from? They can’t just make it up! Once again, the commercial tool is useful in so far as it will record the application instrumentation (assuming your developers DID put some in) on a regular basis and make it available for analysis.
Having read the above, you might think that I am against purchasing commercial tools for performance and capacity within a business. You would be wrong. While there are a number of free tools available, it is right to do a little “what if” analysis on their use. As Capacity Planners, we are used to “what if” analysis. What if the business grows by 20%? What if this new product is highly successful? What if we merge with a competitor?
However, we should also consider the following. What if our monitoring tool fails? With a commercial product that you are paying maintenance for, while the damage will have already been done (you won’t have any monitored data for your systems), you will have some recourse to get the problem fixed. With a freeware solution, you have no recourse to anyone. There may be a network of helpful experts on the internet, but you have no guarantee that they will be there when you need them, or will have a solution for your problem. If your problem is unique and requires changes made to the source, then you are reliant on there being someone that WANTS to make that change and their timescales for doing it.
An initial few hours of downtime could quickly become weeks or months. With a commercial tool, this would not happen.
Freeware monitoring tools have their place. For non-mission critical monitoring they provide the cheapest solution available, but where the Capacity and Performance service depends upon it, I say “commercial” every time.