Dan Blankenhorn at ZDNet has posted some provocative thoughts about the Java CAPS 6.0 SOA suite announcement from Sun Microsystems. His basic take is that Sun fails to live up to its self-generated open source billing. He writes:
A true open source SOA strategy would embrace support for competing alternatives, rather than try to push everyone into paying for (and building) on a Sun-only platform.
True enough, Java CAPS, based largely around the former SeeBeyond ESB, is pretty much all Java platform all the time. I’ve spoken with no small number of people in the SOA space who routinely point out that the Java platform at best is only part of an SOA strategy. Those laments are nothing new. Sun’s approach here is interesting because it’s the opposite of what JBoss is doing. For instance, Sun’s bragging that you get the NetBeans IDE and GlassFish app server with Java CAPS. Yet what if that’s more than you want or need? Maybe you’re not looking for a platform. While JBoss certainly can’t be accused of collecting open source purity points by pushing significant amounts of non-JBoss technology, it is pitching a modular SOA platform.
It gets to the question of how much technology and complexity do you need to pursue service orientation? This is where I repeat the old saw that SOA isn’t something you buy (or download), it’s something you do. Has Sun stuffed too much into Java CAPS or maybe users would be better served to skip the middleware and just use GlassFish? As Blankenhorn points out, in an open source world the app server and service bus ought to focus well beyond each other.
Also, the big SOA-related buzz at JavaOne was around the session on Apache Tuscany. Tuscany is an open source project put together well outside the auspices of the JCP and users at the biggest Java show of the year flocked to it. Apparently there’s healthy demand for open source functionality beyond the Sun platform.
That brings us to the newsiest part of the Java CAPS announcement: Sun is adding MDM tools via a project called Mural. XAware, Talend and Apatar (and others) are already out there offering up open source data integration. Is Mural necessary or does it aim to reinvent the wheel? Eclipse has its Data Tools Project as well. Data integration would seem to be an area where Sun could follow Blankenhorn’s advice and bring some outside technology into the fold.
Sun seems to be stuck in an odd place at the moment where it espouses and embraces many of the laudable benefits of open source software, but it has not yet embraced the concept enough to satisfy the purists or to perhaps even leverage open source to achieve notable innovation.