Scaling the heights of Microsoft Management
Playing with the big boys now
Last week Microsoft hosted the "Best of Microsoft Management summit 2005" at its UK headquarters in Thames Valley Park, marking the first time it has run such an event in the UK.
And very good it was too, with authoritative speakers from the US: Kirill Tatarinov, corporate VP of Microsoft's Windows and Enterprise Management Division, leading development and marketing for Microsoft's management technologies; Michael Emanuel, director, product management for Windows and Enterprise Management; Bill Anderson - lead program manager for the next version but one of Systems Management Server (SMS); and Vlad Joanovic, program manager on the next version of Microsoft Operations Manager (MOM). This was the biggest event Microsoft has hosted at Reading to date and we look forward to more such.
The event was very necessary too, as management of Microsoft systems - once you scale up above departmental level - is still perceived as an issue. Perhaps this should change – Bill Anderson claims that he knows of more users in 100k-plus user sites managed by Microsoft technology than some, allegedly more scalable, solutions have users in total. But enterprise systems managers are cautious and have long memories: for example, Microsoft's Active directory was certainly less manageable than Novell's eDirectory when it was introduced.
So, to some extent Microsoft is playing catch-up with management tool vendors such as BMC (Patrol) and HP (OpenView). It seems to be concentrating on manageability from the ground up rather than from Business Service Management (BSM) downwards.
Microsoft supports the IT Infrastructure Library - (ITIL) for operations “best practice” at last, but it doesn't have a federated, distributed CMDB (Configuration Management Database, a key foundation of ITIL) product yet - in contrast with BMC and Managed Objects, for instance. And of course, the basic concept, of a database of operational systems metadata, has been around since the heyday of the mainframe.
Design of De Times
According to Tatarinov, Microsoft sees itself as adding the “how” to the “what” ITIL prescribes – and, of course, it expects the ITIL community to adopt Microsoft extensions. That is perhaps a little less arrogant than it might once have been once - Emmanuel demonstrated a real understanding of this space when he pointed out that ITIL is about best practice for cost-effective operational management of IT systems but that most of the operational cost has already been locked in by the systems design.
So, you need to design systems for operational managemen and design them to facilitate use of ITIL best practices. The idea of designing and building complete operational business systems - not simply tossing a program over the wall to the business - isn't particularly new, either. But it is a very important idea, not often implemented – and Microsoft is well-placed to encourage it.
Microsoft's approach is built on its Dynamic Systems Initiative (DSI), which is about modelling the complete IT lifecycle and applying end-to-end, standards-based, policy-based management to it.
In practice this means developing Microsoft's System Definition Model (SDM); and promoting the ws-Management standard through the DMTF (), which Tatarinov claims will bring management tool interoperability. (Up to a point, Lord Copper: only freely-available, independent interoperability testing facilities bring true interoperability).
Expect to see "common engineering criteria” management features (such as health indicators and critical alerts) across the Microsoft product range. And, there's a new maturity in Microsoft's product release strategy, with “major releases” (which change the kernel of the product, or Windows) only every four years, and “minor releases” (which should need less testing) every two years – this will make it much easier for Enterprise customers to manage their environment.
A good example of Microsoft's new tools is the System Centre Capacity Planner 2006, based on academic research conducted at Microsoft's Cambridge facility, implemented for the MSN environment internally and now being developed as a general product. This is a simulation tool, which lets you model your systems and “what if” scenarios, visually, out into the future.
It looks like a wonderful tool, although it doesn't yet seem to complete the feedback loop with automatic collection of “actuals” as the organisation moves into the simulated scenarios in real life; neither does it feed back “analysis of variance”, to let you identify assumptions or inputs which have a critical impact on the model outcomes.
Remember that the risk with simulation is that it only reruns past scenarios with different parameters, rather than accommodating radically different approaches and genuinely predicting the future –. But it is early days for this tool, and any form of formal capacity planning will be an improvement on what many organisations manage now.
Part of Microsoft's common engineering criteria approach is providing standardised “management packs” for all of its products, including the tools themselves (although actually implementing these is an ongoing task – don't expect to find them all there yet). These include health indicators, critical event alerts, built-in knowledge (“how to” guides), state monitoring, built in investigatory and repair options and analysis views and reported metrics derived from the product's operational data.
Automatic discovery and deployment seem to feature in this approach, together with role-based management – and third-party, even non-Windows, management packs are on the road map (some are available now), so we are expecting real Enterprise management solutions, not just management of the Windows platform.
Bill Anderson did a very good job of convincing us that Microsoft takes Enterprise systems management seriously and that any past problems with the scalability or usability of, say, Systems Management Server (SMS) are now fixed. However, this implies that there were real issues once, so if you didn't like Microsoft management tools before, now might be the time to take another look (but don't forget that the competition could have improved in the meantime too).
The most interesting presentation came at the end of the day, from Michel Emmanuel, who communicated his systems management vision in a Dynamic Systems Initiative technical drilldown – contrasting “ought-ness” (aspiration) with is-ness (actuals) and applying corrective feedback from “was-ness” (a historical database) and "could-ness" (what-if scenarios).
So, in Systems Theory terms, we seem to have a dynamic (feedback-controlled) equilibrium where a desired operational state can be maintained or moved to a new desired state, in a controlled way. If Microsoft really can deliver on this vision, then it might well become the systems management company of choice. We shall see...®