Nanite is a new way of thinking about building cloud ready web applications. Having a scalable message queueing backend with all the discovery and dynamic load based dispatch that Nanite has is a very scalable way to construct web application backends.
Now to naive old me it sounds like something not really as good as Jini. Its difference appears to be that it dispatches work to "least loaded nodes" rather than a Jini/Javaspaces approach which would have workers determine who was the least loaded and then retrieve the work from the pool based on when they complete their previous task.
Having a scalable message queueing backends isn't new nor is it unusual to use these to do load-balancing. Some people have moved beyond this for cloud computing.
They say that those who don't learn from history are doomed to repeat it, it IT I think we can add the phrase "poorly".
I'm not that depressed because at least it is using a messaging infrastructure for the dispatch and not using a database as a sync point or something else strange. It would be nice though if the number of times we see "new" technologies that were not as complete as those we used in production 8 years ago was less than the number we see that are clearly better. Currently it appears to be 10 "old/new" inventions to every 1 new invention.... at best.
Now maybe I'm wrong about nanite (which would be nice) and it really is better than all of these older technologies and is a brand new way to think about the cloud, it can't just be an agent/dispatcher pattern can it? If it helps people great, if they learn from some of the other projects that have done this sort of thing before then super great, if we get some genuinely new advances around work distribution then fantastic. I just wish projects would not start with the assumption that they are re-defining how people work in a given area without reference to what they have improved upon.
Technorati Tags: SOA, Service Architecture
No comments:
Post a Comment