The implementation of any new technology will cause hardware, software and logistical changes that can affect a...
wide area of resources -- from security and power to vendor relationships and staffing. But while the initial transition to hyper-converged technologies may be complex, it will simplify data center operations in the end.
Don't simply rip and replace
At the physical layer, hyper-converged infrastructure (HCI) is often dense. An older, three-rack legacy platform may shrink to a half rack of HCI. Power and cooling for that half rack will also require careful design and management.
Deploy new hyper-converged technologies before you decommission existing equipment; you may need to run both sets of equipment for week or even months. Implement new HCI nodes into empty rack space and relocate them as you retire and remove legacy hardware.
After you remove legacy equipment, move all of the HCI nodes into sensible locations in your data center to avoid a single point of failure. For example, spread HCI nodes over different power phases and power distribution units and spread networking functions over top-of-rack switches.
Consolidation considerations for hyper-converged technologies
HCI consolidates the storage network onto Ethernet, which forces the storage traffic to share a physical network with the HCI platform management and the VM networks. Most HCI hardware has limited options for networks -- usually two 10 Gb Ethernet (GbE) and two 1 GbE. Limited physical networks must use VLANs to segment traffic. If you are used to many physical networks, make sure that VLANs are setup securely.
A few data centers will need to adopt 10 GbE. Others will simply need more 10 GbE ports for the transition. In this case, release 10 GbE ports as you retire legacy hardware. While mixing 10 GbE port types can cause challenges, some hyper-converged technologies use SFP+ transceivers and allow a choice of cable types. Others will use 10GBase-T with RJ45 connectors. Throughout the deployment process, make sure you have enough switch ports of the right type. Be aware of the out-of-band management ports on the HCI nodes. Generally, these are 10/100 ports, a standard that is generally not supported on 10GBase-T ports. Consider dedicating a physical switch for these management ports.
HCI provides the option to change hypervisors. Nutanix has aggressively communicated the cost savings of using its Acropolis hypervisor in place of VMware's vSphere, and Scale Computing also uses its own hypervisor. Meanwhile, more HCI vendors are starting to support Hyper-V.
If you change hypervisors, you must also change management tools or, at least, the VM management platform. You should also consider changing performance management, backup and any other tools that connect to vCenter.
Reporting and performance data will be discontinued as you deploy hyper-converged technologies. Historic data will be in your legacy platform management tools, and new data will be in the HCI management platform. Your service-level agreement reporting and performance monitoring data will not match up between the products.
You may buy your HCI platform from a new vendor. Many of the main legacy vendors, such as Dell EMC, Cisco and Hewlett Packard Enterprise, now have HCI products. If you choose a new vendor for hyper-converged technologies, follow the usual vendor integration process -- including new contracts and new support processes. Evaluate the different options for new infrastructure and consider how they will integrate with your legacy systems.
A hyper-converged deployment is a large infrastructure change that will also come with organizational changes. Your staff will spend less time working on storage and hypervisors. The network team will remain important, since connectivity is core to IT. Plan a transition to HCI carefully -- particularly if your data center is already full -- and consider the wider impact of the change.
Expect these developments in the HCI vendor market
Large vendors dominate the HCI market
Optimize workloads for HCI technology