|
CEO'S CONNECTION
DATA CENTER AUTOMATION – IT'S A JUNGLE FRAMEWORK OUT THERE
BY ROB GARDOS, CEO
As DCA vendors tout themselves as saviors to an increasingly unmanageable data center, it’s worth taking a step back and understanding what value these solutions actually give you. Demos and canned POCs cobbled together may be enough to drive a decision, but will the solution truly serve as a silver bullet that handles the exponential increase in managed environments? In most cases the answer is no. At best, most DCA initiatives wind up delivering a fraction of what is promised leaving the customer frozen in time often wondering how their ‘automation’ is going to handle the next application upgrade. Why is this the case?
If you look at the fundamental approach to DCA, nearly all vendors focus energies in building the most comprehensive, extensible framework possible. Competition is fierce and they need a story for every environment. As long as you throw enough man-power at a problem you can make a framework do anything. And hey, once the purchase order is cut, then the customer can deal with it. This wonderful ‘framework approach’ results in multi-year implementations that are often out-dated by the time they are completed. Just what we were looking for – an SAP/Oracle ERP project for IT operations – aren’t 70% of those engagements expected to fail?
The truth is that the fundamental concepts of DCA are flawed. While ‘comprehensive’ and ‘extensible’ sound like positive descriptors they really mean unmanageable and unusable. Companies do not need frameworks. They already have C++ and a host of other tools to make their systems do things. Companies need automation content and an ability to easily manipulate that content to conform to their businesses, without scripting. While this ‘content’ approach to DCA will inevitably be less functional, it will actually quickly drive value today and tomorrow. For DCA, usability and maintainability trump functionality (that is if you don’t want to embark on a multi-year disaster with many of the vendors out there). Don’t worry I’m sure they would be happy to give you a great discount on their software just as long as the services engagement is 100 man years of work.
The downside of accepting the truth on DCA is that one product won’t be able to do everything for you. While I don’t love the multi-vendor, best-of-breed approach for any project, it is your only hope to execute on a DCA strategy. Some vendors to think about from a logical stack perspective are:
OS Prov – BladeLogic (BMC), Opsware (HP)
App Management/Config – Solidcore, mValent
Database Management – GridApp Systems
What’s great about the solutions above, assuming you are using them appropriately, is that you benefit from their ongoing suite of content. When Oracle 11g or a new model of patching comes out, you simply can handle this as GridApp adds this capability to their database automation suite. Same goes for BladeLogic/Opsware when RHEL5 is released (good luck using BladeLogic and Opsware to deploy and patch your databases). On the other hand if you are still wed to the framework (i.e. each of these events requires new scripting) you are doomed.
So when it’s all said and done, beware the framework and empty promises from DCA vendors. Functionality is wonderful, but if the solution is not usable or maintainable it’s completely worthless.
Newsletter Home
|
