It must be getting close to Halloween. ZapThink's Jason Bloomberg has recently created somewhat of a tussle in the blogosphere by comparing Oracle's Fusion Architecture to Frankenstein. (Thanks to Radovan Janecek for surfacing this in his Weblog.)
Oracle has “been working on building out their SOA offering for a while by assembling various parts from various different companies," Bloomberg is quoted as saying. "We liken them to the Frankenstein of the SOA world. The question is will it all work when they’re done?”
To which Janecek opines: "SOA is the architecture that supports such approaches. It would be totally useless if it wasn't Frankenstein friendly. Are there really still people believing in the 'Superman' way?"
Zapthink's Ron Schmeltzer also jumped in the fray to further clarify his associate's comments: "The opinion is not that companies shouldn't approach SOA as a Frankenstein of multiple products (in fact, this is actually a BEST practice). Jason and I believe that Oracle's approach itself is a Frankenstein. What this means is that as a company, the company has no single strategy with SOA. Rather than come up with one strategy, they are cobbling together many disparate and unrelated technologies in hopes that it is consistent. For customer to build their own Frankenstein is one thing (and actually a good thing) For companies who have SOA products to build a Frankenstein and then give it to their customers pretending that it's a single product is another thing (and a bad thing)."
Janecek puts it this way: "I would be very suspicious if a vendor told me: 'here is the ESB/SOA fabric from us. Buy it and you will have complete integration platform. You will be OK with this single product and you will not need to incorporate any more third-party infrastructure services (the Frankenstein) into the infrastructure (fabric/ESB) level."
A couple of thoughts here: Too many vendors, even the big ones, are categorizing their products as "SOA" products, which is watering down and confusing the market. Nevertheless, it's perfectly acceptable, and even preferred, for vendors to do their best to tie product lines into a single, coherent vision.
It's more of a sin when perfectly good products are left to die, and not brought into the fold -- to take their place in the next generation of computing. It's incumbent upon vendors with multiple separate product lines to keep all those pieces current and aligned with changes in the market. Each product line has its own constituency that needs to be serviced and kept current. Customers want to know, "What are you doing to make your products work better with X?"
Some say cobbling together multiple products into a single offering is 'a bad thing.' But maybe Frankenstein wasn't such a bad guy, just the product of lousy upbringing.