How much scalability do we really need?
- IT TOPICS:Development, Management, Privacy
Joel on Software has a long, rant-vs.-rant thread on the importance (or lack thereof) of rigorous scalability and industrial-strengthness. I found this courtesy of lesscode.org, who have their own views on the subject.
I find myself in the middle. On the one hand, technology is now mature enough that it's possible to overinvest in high-end tools -- otherwise the LAMP stack, and specifically MySQL, couldn't be such a huge threat to traditional software vendors.
On the other hand, the software vendors know they're threatened. Hence they've been slashing prices and making their products much more useable. And so the incremental cost of building a solid system in many instances is much less than the advocates of quick-and-dirty would like to have one believe.
Remember: A production system is often nothing more than a prototype that was too valuable not to implement immediately.
EDIT: All of the above applies primarily to transactional and other operational systems. Patrick Thibodeau is completely right when he reports that the need for scalability in engineering, scientific, and analytic applications is almost unlimited. That was also the theme of my column arguing the ongoing significance of Moore's Law.

