My Photo

Become a Fan

DailyMile

Google Ad Skyscraper

« Messaging only APIs need to go, JPA needs to step up | Main | Off topic: Switched to Thinkpad Z61p from my T60 »

April 25, 2007

Comments

Bill Higgins

That's interesting. Normally people thing about standard interfaces to allow for varying implementations. You're talking about a given implementation supporting several interfaces!

Billy

Well,
It's only easy to plugin if the standard API is the one they are already using and the problem is there are multiple APIs...

Bill Higgins

I don't see how your comment relates to my comment :-)

Sanjeev Karandikar

Speaking for myself, I would personally give
priority to the following API's specs (at the least so that ObjectGrid is competitive with other offerings in the market)
a) JMS/AMQ - Gigaspaces does have this.
b) Clustered WorkManagers (since Coherence gives this as well).
c) JDBC (I think Isocra's Livestore product does the same thing, except that it is layered over Tangosol Coherence).
d) JPA - I guess if JDBC is implemented over ObjectGrid, then implementing JPA should be trivial enough. I think this gives a competitive advantage over other offerings since neither Gigaspaces/Coherence allows the developer to model in memory caches as relational entities, to the best of my knowledge.
Also Coherence's cache query is more of a simple boolean expression like filter and is not as expressive as SQL/JPA QL (which is inspired by Hibernate QL).
cheers,
Sanjeev.

CH

There is a big difference between interacting with Winamp and with enterprise middleware: that there is only one Winamp (with one API which doesn't change often) and dozens (hundreds??) of middleware stacks (with dozens of APIs which change pretty fast due to user demands).
It is far easier to code a Winamp skin and be sure that it will be relevant in the near future than to satisfy all the enterprise middleware stacks. Standardization could push down the number of middleware APIs, but this is a slow process, usually a few years behind user requirements.
So I guess we are stuck with coding to multiple APIs. What I see different today is that we are acknowledging the problems and the costs associated with coding to multiple APIs. Who knows, maybe we will solve this problem as well..

The comments to this entry are closed.