Simplified IT operations has been the stated goal of the vast majority of data center development for the past three decades or more.
So why is it that the more development we get, the more complicated the environments become?
Virtualization is a classic case in point. Keeping tabs on all desktops, servers and storage units was bad enough in the pre-virtual days, but things became light years more difficult once we added a whole new layer of abstraction to the architecture. And you can expect more of the same once the cloud is up and running.
Naturally, at this stage of the game, selecting the right management system is more guesswork than hard science. With such rapid changes taking place, investing in a technology with a two-year lifecycle could be risky.
For instance, do you follow the path laid out by VMware, which is eyeing more of a service-based platform triggered by the company's planned acquisition of SpringSource? Or do you follow a more traditional systems management approach like IBM's Systems Director VMControl? Or perhaps both?
VMware fans will no doubt be intrigued by the Platform-as-a-Service (PaaS) approach that SpringSource brings to the vSphere cloud system. The company specializes in Web application development and management, offering Java-based tools like the Spring Framework programming model and Apache Tomcat server, as well as the open source Hyperic monitoring system. The plan is to utilize these tools to allow for either internal or external vSphere clouds for application host and management, offering greater scalability and lower costs than current architectures.
For IBM's part, the company has kept its focus on cross-platform management of both physical and virtual systems, with an eye toward handling ever-larger pools of computing resources as enterprises scale up their capabilities. The VMControl system, for example, operates across System z, x86 and Power systems running AIM, Linux and System i, and is compatible with the upcoming eight-core Power7 chip said to be able to accommodate upwards of 1,000 virtual machines.
For those of you just coming to the realization that virtual and cloud environments will require a new approach to management, a good place to start is a new report from Rich Corely, CTO and VP of engineering at software developer Akorri. He points out that while virtualization is good at allocating capacity from a known resource pool, it takes a crack management system to ensure performance is maintained by coordinating activity with both upstream and downstream systems. This is often the first step in breaking down the silos that have dominated IT so far.
What you don't want to do is continue the mish-mash of management systems that have likely evolved in your data center according to Interarbor Solutions analyst Dana Gardner. As his recent interview with three of HP's virtualization management experts points out, a new centralized management approach is better able to detect and prevent problems across the entire infrastructure. This will be crucial as virtual environments make the transition to more mission-critical data.
Some say the most complicated aspect of the modern data center is the centralized management stack. That's probably the reason most managers put off getting one as long as possible. That's an exercise in futility, however, as the longer you wait the more difficult the integration will be.