IBM Joining Liberty

IBM`s joining the Liberty alliance highlights the differences in business model between our two companies.  IBM`s model is the "comprehensive" model.  They support alot of different software standards out there - their value proposition is that if you`re an IBM customer, you can be fairly confident that if there`s a software standard out there, IBM`s software will eventually support it.  The flipside is that it`s typically harder to tie these all together, and IBM is happy to do it for you (either by having you outsource your IT business to them or by helping you put it all together with their consulting arm).

By contrast Microsoft`s model is the "coherence" model.  We are more selective in the kinds of systems we choose to support.  For example Microsoft doesn`t have a J2EE implementation, only support our software on Windows environments, etc.  The value proposition to our customers is that while we don`t support the entire plethora of choice out there, our platform will provide a complete and coherent set of tools with which to go build applications and solutions.  Our promise lays in that we will make it easy to pull it all together.

In some ways this is a difference in mindset over whether IT is a cost-center, which is not core to your business and you should try to let others deal with, or whether it`s a key source of differentiation for your business, a way to make money over your competitors.  Microsoft firmly believes the latter.  That`s why we carefully pick what we`re going to do, to make sure we can provide a simple and coherent platform. 

One important development for Microsoft and IBM is that both companies have come around to strongly believe in interoperability.  Our mutual set of customers is demanding that from us.  So while Microsoft needs to maintain a coherent platform, we are building it on top of an interoperable protocol substrate (WS-*).  Likewise, IBM is adopting WS-* in some part so that they can more easily interoperate with Microsoft software.  In this way our interests are aligned and we`re doing alot of good work together.

So what does all of this have to do with Liberty?  Liberty is a point solution to the Web single sign-on problem, and it may even be a reasonable solution to that problem, but it`s also very limited to its domain (browser-based SSO).  In a generalized solution there are many building blocks that you can abstract out: you want to have the same architectural model for browser-based SSO and rich-client SSO;  the trust layer can be generalized and reused for general token exchange (that`s why we have WS-Trust); you should be able to use reliable messaging in a composable way; the addresses should be EPR`s; etc etc.  That`s why Microsoft and IBM defined WS-Federation - a solution to that SSO problem that sits on top of that generalized WS-* architecture, and use the same building blocks to build protocols for alot of other spaces - EAI, management, devices, grid, p2p, mobile, etc.

Back to the difference between IBM and Microsoft: it`s relatively easy for IBM to join Liberty because that`s their model - do everything.  For Microsoft, it goes against the coherence philosophy - it`s hard to do something coherent when you have to support a bunch of different point solutions for the different application domains.  That`s why we did WS-* - to solve that problem.  It`s meant to be as flexible and applicable as IP and TCP.  And given the broad adoption of WS-*, we believe we`ll get broad interoperability as well.  So we expect to have Microsoft`s identity management/federation software solutions interop well with all of the other important vendors out there, and we don`t believe we need to support or join Liberty to achieve that.