Never say “Never”

I came across this interesting flow chart over at the High Scalability blog.

The chart is quite amusing in a mildly NSFW manner, but what struck us is the context in which it is being used–it is being used in a blog post that discusses whether Twitter’s decision to stick to their current MySQL based storage system for Tweets instead of migrating to Cassandra as their data store is the right decision.

The author of the post, Ian Betteridge, believes it is the right decision–and he uses the above flowchart as one of two props supporting his belief. The other prop Mr Betteridge uses is an old Joel Spolsky blog post which suggests that no company should ever rewrite any of their software from scratch. Now, I am sure that Mr Spolsky is much smarter than I on matters of software, but our experience at Cleartrip is that it’s best to never say “Never.”

At Cleartrip, we have pretty much rewritten from scratch almost all of our core systems–search, transaction processing, transaction management. At some point or other, for some reason or other, we have had no choice but to rewrite these systems from scratch. The reasons for rewriting our systems have varied; sometimes it was for increased reliability and stability and other times it was because we needed a change in architecture to support our growing needs as a business.

We didn’t rewrite code from scratch because we wanted to, we did it because we had to; because it was the sensible thing to do. These weren’t easy decisions to make and there were all sorts of reactions within the team to these decisions–ranging from anguish, resignation, rage and tears to relief.

We knew that investing resources in major rewrites was going to hold us back, but every time we did it, we did because it was the right thing to do for the business. So, while we don’t disagree with most of Mr Spolsky’s position, we will respectfully disagree with any prescription that holds itself as absolutely and perfectly correct regardless of circumstance. Such prescriptions are for political and religious zealots, not for businesses.

Aside

While Mr Spolsky has made many sensible statements on his blog, in this particular case we also disagree with his theory of what killed Netscape. Mr Spolsky believes that Netscape died because they made the “single worst strategic mistake that any software company can make: They decided to rewrite the code from scratch.” Anyone that thinks Netscape died because they rewrote their code from scratch, either:

a) Lives underneath a rock and never heard of a juggernaut called Microsoft that gave away their browser for free, wiping Netscape off the map
b) Is so many sandwiches short of a picnic, that they should immediately cease to be taken seriously in any context

Microsoft gave Internet Explorer away for free, cutting off Netscape’s air supply. Netscape’s revenue stream disappeared–that’s why Netscape died. To suggest otherwise is ludicrous.