Perhaps you saw the announcement at Microsoft Management Summit (MMS) between Microsoft and EMC.
If you missed it, the announcement is here. And some nice press coverage here and here.
But is there more to the picture than meets the eye?
I think this particular announcement carries more weight -- on several levels -- than the ordinary technology mash-up. See if you agree.
So what is this all about?
The Changing Management Landscape
One of my strong beliefs (and EMC's) is that managing the enterprise IT landscape is poised for change at a fundamental level.
In the 1990s, we focused on managing things: servers, routers, arrays, etc.
In this decade, we will need to focus on managing the relationship between things, in addition to the things themselves.
IT has become a pretty complex landscape. Business users now see composite applications that represent dozens or hundreds of entities working together, rather than discrete applications.
And whether it's happening naturally (e.g. IT archaeology, layers on top of layers) or by explicit intent (e.g. SOA, web services), the discrete entity approach we learned in the 1990s is failing dramatically in this decade.
New words and concepts pop up. CMBDs. Correlation. Causality. Root-cause analysis.
All of the sudden, the relationships between things become more important than the things themselves.
Frameworks and repositories wrapped in oodles of custom scripting just don't cut it anymore.
Enter Model-Based Management
Many of us believe a new approach is needed -- model-based management.
The goal is to create a multi-dimensional, usable model of IT that can show different aspects.
How users see things. How the business sees things. How the application guys see things, the database guys, the network guys, the storage guys, and so on.
All different views into the same fundamental model. And all these different views are related, just like it is in the real world. Look at the model differently, see a different relationship.
Ideally, the model would populate automatically, preferrably without agents. And you'd want all your analysis toold to refer the model anytime you have a complex issue where the root cause isn't apparent, or you're thinking about changing something in the environment and want to understand the impact.
That's the approach embodied in EMC Smarts. And that's the approach that Microsoft and EMC are interested in going forward.
[note: when I first wrote this post, I didn't want to drag in the whole SML discussion -- the key standard behind this approach. Well, people figured it out. I guess I'll have to write more on this topic.]
Networks Aren't Separate Anymore
There was one time when the network guys had their tools, the server guys had their tools, the database guys had their tools, and so on.
Network discovery and health monitoring is now integral to server and application management. (As is storage, I would add ;-)
Part of the Microsoft announcement is that Microsoft Systems Center Operations Manager now has EMC Smart's unique ability to do this, and to do this as a natural extension of the Microsoft management model. I think this is a big step forward that I think we'll see other vendors want to take over time.
It Doesn't Make Sense Even For The Big Boys To Go It Alone
Given the paradigm shift in management, I don't think you'll see too many vendors trying to position themselves as be-all, do-all solutions. The technology is very new, there's an incredible legacy of existing tools and methodologies out there, and customers need breadth and depth.
All signs point to an ecosystem model with fundamental knitting in important areas:
- Can customers get to a common CMBD and representational model? Or the ability to usefully federate multiples?
- Can existing tools and newer specialized tools feed into the new modeled environment to create a larger picture?
- Will there be useful standards around how capabilities and causality is represented, enabling the tools to work together?
- And, once the analytics make a recommendation, can downstream workflow be driven for problem remediation, etc.?
The real action will be at the representational model that all can feed, and all can exploit. In my mind, it's all about the model.
I'd like to think that EMC Smarts has a good headstart in this direction, but I'm sure the footrace is far from over.
And if you'd like to read more on the topic, head over to Josh Maher's post on this.
And This Topic Should Heat Up In The Future
The forcing function for a new approach isn't going to let up anytime soon.
IT is getting more complex.
There are more pieces, and fewer people. Users are increasingly intolerant with poor service delivery.
Applications are now inherently more complex entities, with SOA adding fuel to the fire.
It doesn't look like it's going to let up anytime soon, does it?
Comments