It's no secret that IT managers are under pressure to increase service levels while decreasing operational costs. But over the past decade, data centers have grown in complexity to support business demands, so addressing the cost/service challenge has become ever more complex. To that end, companies are increasingly turning to software-based productivity tools to manage their operations.

Such software is big business. In July 2007, HP paid $1.6B to acquire Opsware and in March 2008, BMC acquired Bladelogic for $800M. But despite the more than $2B changing hands, and the major companies involved, surprisingly few members of the IT industry can pinpoint exactly what Opsware and Bladelogic do.

Can you?

The answer is that both provide aspects of data center automation software: software designed to increase operational efficiency in the data center. But what exactly is data center automation, where does it fit into my data center, and what's still missing?

Simply put, data center automation is the ability to replace manual processes by automating tasks such as provisioning, patching, and software distribution on a single server or multiple machines. Important functionality includes configuration management and the ability to model complex services and application environments.

Examining the existing offerings, we can see that they're not off to a bad start. The technology certainly beats manual process for tasks such as patch management on multiple machines. But most offerings are based on custom scripts, and all the scripts are still statically tied to the underlying infrastructure. The result: two significant problems emerge.

The first problem is that of (in)flexibility. Many of the automation providers show a surprising inability to handle changes in the IT environment. If you're using data center automation to automate tasks, then you have to capture all the actions that fully correspond to the particular configuration or fault recovery event. But what if one of the steps requires configuration changes to the server, OS, network, or storage? Or what if an IT administrator changes a switch port, IP address configuration on the subnet, network booting, power management tool, or WWN? Unfortunately, you'd have to address any of the changes manually. If you run an old automation script in an environment that has changed, you'll actually create some unintended consequences. So, data center automation is dependent on the current and the expected static state of your infrastructure.

Scripting is a DIY approach to automating static elements in your data center. Scripting is great for automating a single problem once. But, your IT administrators would have to determine the state of the data center and also set up the logic for changes in the environment. The result is that your IT administrators have to become programmers. It's not efficient for IT staff to become programmers when what you really want is a software solution.

The second problem is the need for component administration. Your company still needs to manage every component for every vendor in the IT environment and change scripts whenever the company makes a new vendor choice. If the script is set up to do remote management on HP blades today and tomorrow you buy Dell blades, then you need to change the scripts. There's no abstraction between the physical hardware layer and the scripts. This results in more manual script configuration to add to the many manual processes already in place.

So what's the missing component needed to successfully deploy and meld the current offerings into flexible, adaptive, real-time data center automation? The answer is Infrastructure Virtualization.

As the older automation space is consolidating, the infrastructure virtualization space is emerging. The leading candidates include software from Scalent Systems V/OE (Virtual Operating Environment) or Unisys's uAdapt. Infrastructure Virtualization solutions are geared to supporting the existing systems deployed in the data center, this includes x86 and SPARC servers running Windows, Linux, and Solaris on bare metal as well as VMware ESX, Xen, and Microsoft hypervisors.

The upside of Infrstructure Virtualization is that it takes a bare metal environment, manages it, and publishes a consistent view regardless of the make and model of the underlying servers, switches, and storage arrays so that companies can achieve a truly dynamic data center. This is data center nirvana because changes can be made without modifying preconfigured scripts due to the abstraction layer these technologies leverage and present.
Infrastructure Virtualization software also provides increased operational efficiency and flexibility for repurposing physical or virtual machines. IT administrators can move server images between physical and virtual machines and back again. Sometimes it may be desirable to run an image on its own bare metal, but other times, it may be preferable to run the image within a virtual machine. Infrastructure virtualization software facilitates these changes without the need for physical intervention.

The key point to remember is that many of the tools that purport to do data center automation are geared to static environments. There's a big difference between a scripted approach for static data centers and an abstraction layer approach to managing dynamic data centers. So, check out real-time data center automation solutions. Research Egenera, test out Unisys uAdapt. Give Scalent a call to learn more about how to gain scalability and efficiency by creating a dynamic IT environment. Your data center is ready - are you?

Alana Achterkirchen, Director of Marketing, May 2008