Ivelin Radkov - Fotolia

Sharpen your server management and monitoring strategy

Effective server management requires monitoring, documentation and, sometimes, a move off premises. Discover top server management tips and how to apply them to your data center.

Server management and monitoring is an essential part of data center administration. It requires a holistic view of the servers you'll need to manage, a thought-out monitoring strategy and careful preparation of what to do once your servers stop working. And, in an industry saturated by cloud and colocation, server management increasingly requires a plan to move equipment off premises.

Here are tips to hone your server management and monitoring strategy, taking the needs of your IT environment into consideration.

Don't let documentation be tedious

To effectively manage your servers, you'll need proper documentation. Collect as much information as possible about your server, from serial number and model to SCSI details and interface type. If it's a virtualized server, you'll need to include virtualization vendor, hypervisor product and version in the inventory.

This process can be extremely tedious if done by hand, but you can use custom scripts, such as those in Windows Server PowerShell, to automatically capture server configurations and inventories. Implement a comprehensive change management strategy to prevent making any system changes that cause accidental disruptions to other systems. For example, Microsoft's desired state configuration ensures that each hardware or software component doesn't change from its known state.

Monitoring is key to prevent resource waste

Although consolidation and virtualization have decreased space issues in today's data centers, 30% of servers are considered comatose or "zombie servers," according to the Uptime Institute. These servers suck expensive resources without offering any usefulness. Either directly or indirectly, zombie servers unnecessarily hog power and cooling energy.

Careful server management and monitoring is crucial to combat zombie systems. First, choose either a data center infrastructure management (DCIM) platform, a configuration management database or a homemade automated tool. With the tool of your choosing, monitor CPU utilization and power draw, and then identify underused resources to spot and eliminate zombie servers. An alternative option is to assign a staff member to regularly measure and benchmark server utilization rates.

Upgrade or replace?

A critical part of any server monitoring and management strategy is the decision of whether to upgrade or completely replace aging systems. It's usually more cost-effective to simply upgrade, but there are other factors to consider as well, such as growth curve and performance. If your memory or storage upgrade doesn't last until your next budget cycle, it can ultimately be more costly for your organization.

If your server can accommodate a solid-state drive, moving it from a spinning disk drive can dramatically increase performance. But this upgrade can cause performance concerns, as well -- improvement to one part of the system may force other parts to try to keep up, which can create bottlenecks.

When it comes to an upgrade, the type of server makes a difference. Blade servers are accompanied by strong vendor lock-in, which makes the addition of commercial off-the-shelf (COTS) components difficult. And sometimes, it's not an option to upgrade a blade server because the vendors may not continue making that specific blade family.

Rack servers, on the other hand, are easier to upgrade. A rack server has a lifespan of 3 to 4 years, which falls into the refresh cycles of most organizations. In addition, the option to add COTS components is generally more likely with rack servers, which can make upgrades easier.

Moving prep done right

At some point during your server management and monitoring process, you may need to plan for a move from in-house to a colocation facility -- a process that requires a lot of preparation. Calculate the power, space and cooling that you'll need for the servers that are about to move. Since many colo facilities have their own pre-wired racks, you may need to distribute servers differently and create a new physical map for your equipment.

Determine whether you can afford downtime during the move. If you can, turn off servers dedicated to a specific workload, move them and turn them back on. If not, use virtualized servers to run jobs on a subset of the systems so you can partially shut down servers during the move. It's helpful to appoint a "move manager" to identify and communicate any issues that arise.

On the day of the move, shut down or move apps to other virtual machines before you retire the servers. Tie or tape down the internal and fiber-optic cables to keep them in place. It may be worthwhile to hire a computer-moving company, but either way, you should use an air-cushioned van to transport the servers and avoid bumpy roads on the ride. After the mover delivers and installs the servers, connect the power, cooling systems, switches and routers -- and then do a visual check to make sure that nothing is loose or obtrusive.

Next Steps

Learn about the lock-in risks of blade servers

Explore your options for remote server management and monitoring

Discover tips for IT ops automation

Dig Deeper on Server hardware strategy