During the Redmonk unconference at CommunityOne, there was a discussion about whether REST was going to kill Web services (WS*). This discussion followed a discussion on OSS and how ‘different’ OSS really was from the Traditional software model at the end of the day.
I’d read a good deal about REST being the death blow to WS*, but didn’t know much else, so, I was looking forward to learning more.
During the REST/WS* discussion, folks pointed out that Amazon processes about 80% of RESTful service calls (from 3rd parties), with WS* making up the remainder. Then someone from eBay mentioned that 80% of 3rd party service calls are Web service calls, with REST making up the remainder. Would those percentages change? Maybe, would they become 100% to 0%? Unlikely. There were things that REST & WS* were each much better than the other at, and vice versa. For example, if you just want data and the service provider isn’t wholly concerned about how you use that data, REST is probably a good model. If security matters, WS* is likely the best practice. etc., etc., etc.
The big “ah-ha” I took away from the two discussions was more evidence against the “there can be only one right answer” line of thinking that seems to be widespread amongst OSS proponents.
One size fits all is a great motto for a sock vendor. Not so much for a software vendor; or the software vendor’s customers!
PS: Thanks to the Redmonk guys (Cote, James & Stephen) for keeping us engaged in the discussion.
PPS: I’m going away for 2 weeks, so my next entry will likely be about how freakin’ cool it was to walk (or in my case, be carried along) the Inca Trail