Jeremy Wright posted an entry the other day titled, “Web 2.0 Companies Need To Scale” where he talked about how infrastructure scalability is sometimes not written into the plans of new “web 2.0” companies and I agree. He then continues to talk about how executives at companies want 3 or 5 9’s of uptime (99.999% uptime) but aren’t willing, or aren’t aware of, the massive cost at each step of the way. I thought his article definitely needed to be written, and kudos goes out to Jeremy for having the foresight and cojones to actually write it! Oddly though, that article has been under fire for a day or so now and I think some of what Jeremy is saying has been misinterpreted.
Jeremy’s article was analyzed yesterday by an engineer who runs Feedlounge and he was probably not too happy to be targeted by Jeremy’s entry directly. Jeremy talks about how some “web 2.0” companies and services like Feedster, Technorati, Delicious, Google Analytics, etc. have been wrought with scalability and server downtime issues and I’d have to agree. Paul wrote about Technorati’s issues a few months back, but now they addressed them all and I use Technorati more than I use Google.
The article was given even more gasoline to burn on when the Gigga Man himself, Mr. Om Malik, wrote up his own thoughts on the matter which actually coincide with what Jeremy was talking about. Business plans need to think about their infrastructure if they plan on supporting lots of users. Companies who don’t think about scalability are probably going to be suffering the consequences. All true stuff. I agree. Well, David from 37signals vehemently disagreed.
David interpreted Jeremy’s thoughts as, “all web 2.0 companies need to have 5 9’s of uptime in order to be successful” and then went off on a pretty good rant about how planning for what you’re at right now is a better idea than purchasing 7 figures worth of load balancing equipment before you’re got a working prototype. I agree with David’s point about not wasting money on things you don’t need yet, but I truly don’t think Jeremy intended his entry to be a Ten Commandments of Web 2.0 Scalability. He merely pointed out that some companies have been having downtime issues, and that they should have been thinking about scalability right as they saw the issues pop up. I don’t think he was insisting on spending ludicrous amounts of money on mesh right away, but maybe I’m mistaken.
Liberalcowboy at Jack Of All Blogs picked up on all the hullabaloo and mocked some of the B5 guys, and then in the comment section all hell broke loose as the B5 guys, Liberalcowboy, Scrivs, and a bunch of other people started voicing their opinions about lots of stuff. Crazy times.
I’m sure Jeremy never imagined his seemingly innocent entry would get around as much as it did, but that’s the weird thing about the blog world. Things you don’t think matter probably matter to somebody, and the entries you spend a lot of time on might not get any comments at all lol. Go figure!