Over the past two weeks I've worked on and off at building out our premier replacement web server and slow but surely prepping it for testing trials at the end of the month. Two things strike me about the process. First no matter how well documented the prior build out was you inevitably encounter unknown elements that must be fully vetted and justified. Upgrades are an inevitable part of the roll out processes and chances are the install docs for an application's dependency have become outdated or simply aren't applicable any longer.

The second truth of hardware building is that while it's as much art as science success has more to do with measured steps and exactitude than anything else. I once again find myself comparing the process to auto racing in that building a good server is like building a good race car engine. Performance and reliability are your greatest concern but these two qualities can often be diametrically opposed. Only through diligent research, thoroughly testing and precision installation can you ever hope to achieve and acceptable balance of both under load that you expect.

Of course race engines are not subjected to penetration testing, scalability and other IT centric requirements. Still it's important to craft your mighty server with all the care of a master engine builder. In many ways the server is the engine that drives your application and serves the ultimate goal of keeping your clients happy.