New servers are typically offered with a bewildering array of processing, memory, storage, connectivity and other capabilities. But tight budgets often mean that during the initial procurement of your machines, these features have to be kept at a bare minimum. With the economy picking up, perhaps you have some slack in your budget or have been able to save money in other areas of the business. If you are considering a server upgrade,...
how can you ensure you get the most out of your investment?
When looking at a server upgrade, it’s important to understand that we're entering a period of time where capacity is scaling out, not up. Generally, capacity problems occur not because workloads become more intense and require more computing capability on a single machine, but because workloads need to spin up additional nodes to crunch tasks. In other words, the software being more capable isn’t a problem; more capacity is needed to run more instances. This issue frequently occurs with cloud computing technologies. Therefore, a server upgrade can make financial sense and prove to be a worthy venture.
Here are four areas where cash spent on existing machines can enable a healthy return on your server upgrade.
- Memory: For the best performance-per-dollar improvement, add memory during a server upgrade. The importance of memory is increasing as virtualization becomes more popular. Suddenly, vast swaths of system memory are committed to a hypervisor that you were previously only using during peak workloads. Spending some money on increasing memory to the upper limits of a system’s capabilities actually saves money by enabling consolidation through virtualization. In turn, response times for workloads are improved on a taxed system. A memory upgrade increases system capacity, takes advantage of hardware that’s already been purchased and extends the useful life of a server system by two to three years.
- Storage area network or fast storage: If you’re lucky enough to have a budget for a server upgrade, consider spending some of it on faster local storage for your servers. Drives that spin at a higher RPM to reduce head-seek time and disks that use a higher-speed throughput protocol will improve system performance. Aside from the performance benefits, upgrading your disk storage subsystem can also reduce power consumption, as today’s disks are manufactured with low-power goals in mind. When low-power disks are paired with a controller and an operating system that supports reduced power profiles, energy consumption is dramatically lowered and there is a quick payback on the new, faster and cheaper-to-operate disks.
- Additional network controller: Again, as virtualization becomes increasingly popular within data centers, the advantage of having multiple network controllers becomes important. From a redundancy perspective, the ability to bind separate physical adapters into one logical connection is helpful. From a functionality perspective, having hard connections to two distinct networks can make configuration and administration easier. This upgrade may be more difficult if you are working with rackmounted servers, as there is often limited expansion space for additional PCI- and PCI-Express-based components within the actual motherboard and chassis. However, if working with mid-size or even minitower machines that are racked differently, you may have the space to consider adding another Ethernet card.
- Backup or disaster recovery features: If there is budget to spare in a server upgrade, consider investing in spares. Disaster recovery solutions are expensive and sometimes get left out of the initial procurement process, but they are, at times, critical features of data center infrastructure. Don’t tempt fate. Remote access cards that allow you to remotely restart, power off, access the console and do other tasks that ordinarily require you to be physically present at the server console are expensive but well worth it in the event of a problem. Consider a RAID setup for file and mail servers, or a more expansive array, to future-proof capacity and disaster preparedness in one fell swoop. Don’t forget about power needs too. Buying uninterruptible power supplies, redundant internal power supplies and hot spares for system equipment can also be a good use of funds.
ABOUT THE AUTHOR: Jonathan Hassell is an author, consultant and speaker residing in Charlotte, N.C. His books include RADIUS, Hardening Windows and, most recently, Windows Vista: Beyond the Manual.
Dig deeper on Data center budget considerations