Data Center

Moving to Hyperconvergence? First, Know Your Needs.

Here's what agencies need to consider before adopting hyperconverged infrastructure.

Rick Cook learned programming on a computer with magnetic drum memory. Since then he's written thousands of articles on all aspects of computers and high technology — as well as several fantasy novels full of bad computer jokes.

Agencies see a way to reap the benefits of the cloud and to ease the strain of managing additional systems: adopt hyperconverged infrastructures.

Hyperconverged systems combine storage, networking and processing in a single appliance and offer a series of advantages. Notably, these systems often are easier to expand and can accelerate application delivery. Better yet, they can save money in hardware, support and maintenance, making them prime candidates to deliver desktops from a virtual machine on a server.

Here’s what agencies should consider before adding a hyperconverged system to their infrastructure:

1. Are Accelerated Graphics Necessary?

Advanced graphics capabilities are essential to some functions, such as data visualization and high- performance computing. These are key to applications related to geological exploration and video processing. Hyperconverged systems vary in how well they support these capabilities, and some systems do not support processing units for 3D and accelerated graphics. If these features are critical, make sure the system can support graphics processors at the appropriate level. Otherwise some functions may not work.

2. Know Which Applications You’ll Use

Although many hyperconverged systems share features, such as a basic architecture and the use of modules or nodes for expansion, not all are alike. The appliances do not feature standard components, and systems vary greatly in the hypervisors they support. In addition, some hyperconverged appliances are designed for specific applications. HPE and Cisco each offer appliances built with virtualization in mind. This makes it critical to understand the applications you expect to use at the beginning of your search.

3. Calculate Your Workload

Although IT staff cannot optimize hyperconverged systems for performance to the same degree as conventional systems, they need to understand the characteristics of their workload from the outset. By taking advantage of systems’ monitoring and logging functions, IT teams can determine their requirements for bandwidth, memory and storage capacity. Once a system that provides these requirements is in place, performance tuning is largely limited to adding more RAM. By selecting a hyperconverged system that meets the demands of a specific workload, IT teams can avoid problems such as bottlenecks.

Lance Pettiford