Pattern-Oriented Software Architecture Volume 3: Patterns for Resource Management

Foreword by Steve Vinoski

If you've been a software developer for any appreciable length of time, you have almost certainly experienced what I call the ‘stroll down computer memory lane’. This event occurs regularly when developers get together in a social setting, such as having lunch together. It starts out innocently enough, with one of the developers describing a recent run-in with an especially difficult problem. Not to be outdone, another developer chimes in with a story detailing the conquest of an even worse problem. From there, each new story attempts to outdo the last, until only the old-timers are left speaking about primitive machines that had to be programmed with punch cards or toggle switches and had only a few bytes of RAM. I am waiting for the day that during a stroll down computer memory lane, someone tries to convince me that back when he or she started programming, there were only zeros to program with, and no ones!

Developers are able to compare stories in the manner described above because programming inherently requires many trade-offs. Applications have to share computing resources with other applications. Memory space and disk storage are not infinite. CPUs can process only a certain number of instructions per second. Disk and device I/O can take a relatively long time. Establishing database connections and network connections can be expensive in terms of time and resources. Throughout the history of electronic computing, tremendous advances have been made ...