Hyper-convergence makes it easier for IT administrators to stand up infrastructure for specific workloads because the hardware and software are so tightly integrated. But that can result in resource shortages or surpluses that don't make the best use of snug IT budgets.
Converged and hyper-converged infrastructure (HCI) are still fairly young, and they have room to evolve and grow. Companies don't yet know whether vendors will be able to offer a solution to the problem of expanding a stack of hyper-converged hardware. And hypervisors still have memory and density limitations that can be concerns for some shops.
But a move to more modular and powerful hardware topped one expert's predictions when we asked, "What changes do you expect to see from converged and hyper-converged infrastructure in 2017?"
Brien Posey: More than likely the HCI market will continue to grow. Gartner predicts that the HCI market will grow to almost $5 billion dollars by 2019, and will make up 24% of the market.
I am inclined to agree with Gartner regarding the continued, steady growth of the HCI market, so long as the economy does not drastically change. The value of the dollar could conceivably decrease, resulting in inflation. Depending on the severity of this monetary adjustment, some companies could decide to hold off on investing in hyper-converged hardware or software until the market rebounds, which could dramatically slow the projected growth of the HCI market.
Financials aside, I also predict HCI vendors will focus on improving HCI scalability. Although HCI was once primarily geared toward smaller organizations, hyper-converged hardware is becoming ever more powerful. The availability of powerful hardware certainly helps with workload scalability, but companies must still consider hypervisor limitations.
My guess is that VMware and Microsoft are each working to make changes to some of the limits that are inherent in their respective hypervisors. Specifically, there are five areas where I expect the limits to change:
- Total memory supported per host
- Total memory supported per virtual machine (VM)
- Maximum concurrent running VMs per host
- Maximum concurrent running VMs per cluster
- Maximum number of nodes per cluster
Both VMware and Microsoft have adjusted these limits in the past as technology has improved, and I do not expect the current hypervisor limitations to remain in effect indefinitely.
One more change that I expect to occur over the next year to two years is HCI becoming more like converged infrastructure. Today, hyper-converged hardware nodes include built-in compute, storage and network resources. Having fully integrated nodes helps simplify deployment, but it also unnecessarily increases costs in some situations. If an organization needs more compute resources, it must purchase nodes that also include storage resources. Those storage resources come at a cost. This can translate to wasted spending if the storage resources are not needed.
I predict that HCI vendors will eventually separate nodes into standardized compute modules and storage modules. HCI customers will likely be able to purchase the module combination that they require, and insert those hyper-converged hardware modules into a chassis.
Choosing the right hyper-converged storage
How to buy HCI
Who rules the hyper-convergence market?