Measure Twice Cut Once
It’s pretty noticeable when this isn’t done in the real world. Software being, well, soft means not having to worry about such things as not fitting or being made of the wrong material or too big to get through a standard doorway. They still inflate the costs, both in the initial development and in maintenance.
The business side generally understands that, like real books, there’s no one right way of writing software, with vastly different approaches resulting in comparable functionality. The business side also knows that recognizing tight, efficient and extensible code is difficult and, more to the point, beyond their direct ken. That’s what technical managers are for. Whether bloat or turgid, the bottom line is that this uncertainty is accepted as a cost of doing business, with varying degrees of tolerating the situation.
The source of this is uncertainly can be materially reduced if there were some objective metrics on how a given piece of software is doing. With this information, realistic and accurate comparisons of befores and afters can easily be done, against which the relative costs (e.g., programming and ancillary labor) can be determined.
There are quite a few ways these metrics can be obtained, with surprisingly little effort. The most obvious, and essentially free, is examining the desktop and/or server event logs. The server logs themselves can provide insightful information as well. Web sites can also be measured effectively, for such things as content size and loading, and of course tag errors that may mask other problems (e.g., sloppy logic).
Software being developed can be similarly measured, eliminating the tendency for new code to expand like the San Jose’s Winchester House. For both new development and maintenance, effectiveness, as well as stability and reliability, can be checked by running the existing code in parallel with the changed code. This has become dirt simple with offerings like Amazon’s AWS EC2 services, where complete data centers can be rented by the hour, set up and torn down without the commitment of building captive infrastructure, and not requiring in house system administrators to spin it up and down.
By far, the initial investment in software is dwarfed by the cost of maintaining it. Finding out what has to be changed and the effort required before hand, along with consistent and objective metrics, can make that investment’s returns something that you can take to the bank.
© Copyright 2009 Chuck Brooks for FutureWare SCG
A Word From Our Sponser
Finally, a fast way to fix 
html scripts without having to rebuild them!
Tags: accurate pricing, amazon aws ec2, reducing TCO, software maintenance