WavebreakMediaMicro - Fotolia
There is no single, universal methodology for introducing hyper-convergence into an enterprise. The exact approach and timetable will always vary based on each organization's specific needs or goals. However, the general consensus seems to emphasize phasing in hyper-convergence.
A traditional, manually integrated data center storage, networking and server setup is a complex entity that represents a huge financial and labor investment for the organization. There is almost no justification to forklift an entire data center worth of IT equipment in favor of hyper-converged gear at the same time -- the expense and disruption is simply unacceptable. Instead, a hyper-converged infrastructure setup typically joins the existing enterprise environment to serve a specific project or technical requirement.
There are good reasons for a hyper-converged setup to exist simultaneously with traditional IT infrastructure. An organization makes a minimum investment, just purchasing the initial deployment. Administrators can systematically deploy new workloads, and have ample time to gain experience working with the hyper-converged system.
Once the initial Hyper-converged infrastructure deployment is complete, an enterprise can expand its setup over time by adding more nodes. Each node includes compute, storage and network resources under a common management platform, and integrates seamlessly into the product. Administrators can also use their experience with hyper-convergence to plan new deployments or migrate existing workloads from the heterogeneous infrastructure onto the new hyper-converged infrastructure.
Migration initiatives are most common at budget or refresh times in a company's business cycle. Servers, storage arrays and other data center equipment doesn't last forever, and the ongoing need to replace aging systems can provide IT leaders with an opportunity to transition more to the hyper-converged infrastructure. This strategy is effective in a virtualized data center, where the IT team can migrate VMs onto new hyper-converged nodes instead of new servers.
Migrations need not be driven solely by refresh cycles. Some organizations make the decision to transition less-critical workloads immediately. Once they have a process in place that streamlines and protects the migrations, the team can move the most mission-critical workloads onto the hyper-converged infrastructure setup. This gradual approach using a process well-refined by previous experience minimizes risk and mitigates potential disruption.
Do your infrastructure vendors stay the same?
Some hyper-converged infrastructure products come from names familiar in the heterogeneous data center space, but many are specialists.
VMware introduced a hyper-converged reference architecture called EVO:RAIL, which includes a comprehensive software stack and hardware requirements for implementation. Partner vendors like Dell, EMC, Super Micro and NetApp manufacture the hardware, integrate VMware's stack, and provide the service and support.
There are several examples of vendors primarily in the hyper-convergence space. SimpliVity Corporation offers the turnkey OmniCube family powered by its OmniStack software stack. Nutanix offers its Acropolis storage and compute fabric along with its Prism management software. Nimboxx hyper-converged infrastructure appliances offer servers, storage and networking with unified management. Maxta Inc., provides the MaxDeploy appliance along with MxSP software to operate the system.
New vendors are always entering the hyper-convergence market with software stacks designed to facilitate easier management and software-defined infrastructure capabilities. Considering the speed at which new hyper-convergence setups appear and evolve, always test prospective products carefully through extended proof-of-principle projects before committing to an actual purchase and deployment. Once you contract a vendor, it can be difficult to later migrate to other hyper-converged platforms. Careful investigation now avoids tremendous disruption and wasted capital to change.
Six places to find data center savings
What's the best for on-premises to cloud migration?