Server Virtualization's Next Step

With server virtualization rollouts well underway in IT data centers throughout the country, IT managers need to begin considering how best to augment these deployments. Server repurposing is key to making the most of a virtualized environment.

Server repurposing is the ability to change such things as what servers are powered on, what operating system those servers are running, and how and what networks they are connected to. All of these activities can be a manual ordeal.

Software applications are available for server repurposing. Examples include Scalent Systems V/OE 2.0 or Unisys's uAdapt. You can also use proprietary hardware-based solutions from Egenera.

Simply put, server repurposing software virtualizes all of the connections that would normally require manual interaction and allows you to repurpose a server at the speed of boot time as opposed to the hours or, more likely, days – it would otherwise take to reload software, change configurations, or move machines and cables.

Use cases

In an environment that has deployed or is planning to deploy server virtualization, server-repurposing software is a vital component in a successful rollout for ongoing operation of the virtualized environment. With server repurposing in place, the movement of servers and applications from a physical environment to a virtual environment – or even more impressively, from a virtual server environment to a physical environment – is made possible. This provides important capabilities during a server virtualization rollout.For example, if during this rollout you find an application that either does not behave well in a virtualized environment or simply consumes too much of the physical server's overall resources, then with a server repurposing tool you can very easily move this server from the virtual environment to its own physical server in as much time as it takes to boot that physical server.

Another case may be if you need to perform maintenance on a server that is actually running the server virtualization operating system. Downing this server now also means bringing down all the virtual servers it is hosting. Most server virtualization operating systems have some form of a High Availability (HA) solution, but this requires that another server of similar power and configuration be powered on, running the same virtual server operating system and connected to the same network. This requires additional power and cooling and, of course, the HA components of the virtualization operating system.

With server repurposing software in place, maintenance of an active server virtualization host is greatly simplified. Instead of needing to have the HA software running or suffer downing all the virtualized hosts, you can power on a cold server, load the virtualization software onto it, and start all the virtual servers and make all the right network and storage connections within in a matter of minutes.

This solution is significantly more cost effective and far more power efficient than the manual alternative. What's more, the second server need not be an idle machine, powered off. Instead of waiting for a downed primary server, you could be deploying the second server in a test environment or for some other function. Server repurposing allows the flexibility to quickly switch this server's role to a primary server hosting multiple virtual machines.

Server repurposing not only adds value in virtualized environments, it also enhances non-virtualized environments. The ability to re-deploy a server to a cold server in just minutes gives you a form of near clustering with far simpler deployment. In addition, migration to newer, more powerful servers is as simple as plugging the server into the network and storage and having the server repurposing software take over from there.Lastly, in a lab environment, server repurposing plays a critical role. The ability to have a test environment that can change from one test bed to another and back again can make a significant reduction in lab expenditures as well as provide great productivity out of the lab environment.

A typical test plan will hit pauses. If one has the ability to redesign the test environment quickly, other testing could be started or continued while results from the prior tests are being analyzed. Then if more testing from the prior test needs to resume, the lab can quickly switch back to the original configuration.

There are other solutions that restore images to a server, but a server repurposing tool manages all the connectivity as well, which is critical for speed and flexibility. Without the ability to manage the network and storage connections, the manual process of re-cabling and reconfiguring will typically be painful enough that you won’t switch to a new server.

So in the situation just cited, without server repurposing, a decision or compromise has to be made between slowing down the testing process or purchasing and powering additional servers to run both tests simultaneously.

Also, in a lab environment that is complimented by server repurposing, you can quickly redeploy servers into the production environment in case of emergency needs. While this can be done without server repurposing, most labs are deployed on separate networks attached to separate storage and therefore adding servers requires manual re-cabling and physical movement of hardware. Because of the amount of effort required to move a lab server into production, it rarely comes back to the lab, or because of these challenges, a new production server is purchased. With server repurposing, the process can all be done without any physical re-cabling or movement of server hardware.Components of server repurposing

What are the components of a server repurposing solution?

Typically, there is a controller server. The server repurposing software is installed on this server and it logically sits on the “side” of your network. The controller software manages the physical and virtual hardware, software, and network configurations. The controller server must have Layer 2 access to the switches and networks of servers to be managed.

The controller also hosts the management console or user interface to the controller. This is used to configure and monitor the physical and virtual elements in a data center.

In some instances, agent software is automatically installed on every server image to be managed. The agent software communicates with the controller to obtain configuration information and provide status information. A heartbeat is maintained between the controller and agent software. If the heartbeat to a server fails, the controller will select another suitable physical or virtual machine to run that server.

Notably, the agent is typically not in the server's data path – it is present purely for management reasons. So there is little added overhead or additional risk from the possibility of agent failure.Server repurposing and data center virtualization go hand in hand. Server repurposing, via whatever product you choose, allows for complete optimization of an investment in server virtualization and should be considered in all rollouts of the technology.