4 ms·
> Pre-Emptive Scaling Man, this one a thousand times over. I shudder when I think of all the work that could have been saved if these premature optimizations h
by derwiki 14y ago
> Pre-Emptive Scaling
Man, this one a thousand times over. I shudder when I think of all the work that could have been saved if these premature optimizations had never been made (I wrote a blog post on 500m row tables in MySQL about this). A little dose of YAGNI would go a long way to many startups. Also, re: Twitter and Tumblr's scaling problems: they didn't fail because of sporadic downtime. They fixed the actual problems/bottlenecks when they encountered them, and went on.
- ftwinnovations 14y agoAgree 10,000x over your 1000x. On the occasions where rapid unexpected (and awesome!) traffic or growth spikes happen you can almost always temporarily patch the situation by scaling vertically (bigger/faster disk/CPU) or quick and dirty horizontally (load balancing or replication) and then have some breathing room to work out the real meaningful or cost effective scaling solution.