The SML debate with Piyush continues. He has responded to my earlier post. Piyush agrees that SML is a robust specification but wonders why one should care about it. He points out that deployment-related problems should be solved through platform simplification rather than by inventing tools such as SML. Finally, he is sceptical that SML will get any traction outside of Microsoft products.
I agree with Piyush that platforms, IT services, and software products should be designed to be manageable. Unfortunately, we are not there as yet. Right now, manageability is probably at the bottom of the list when it comes to prioritizing and building features for products. Model-based management, the key value proposition of SML, remains relevant even for products that are designed to be highly manageable. Why? Because there will always be considerable variance in the deployment environment and configuration of an IT service or system since a "one size fits all" approach is impractical - organizations vary in size and georaphic spread, have different hardware and OS, have different IT policies, are subject to different regulations etc. SML models add considerable value in such scenarios by allowing ISV, SIs, consultants, or IT administrators to define models that capture the configuration and policies relevant to an organization, and using these models to monitor compliance and/or enforce them.
Will SML be adopted by other vendors besides Microsoft? Only the passage of time can answer this for sure. The SML WG has 11 members, and most members have publicly announced their intent to support SML in their products.