One of the initial appeals of hyper-converged infrastructure is its use of commodity hardware which in theory lowers costs. The reality is that most commodity hardware isn’t able to fully utilize high-performance CPUs, internal connectivity and high-performance NVMe media, which leads to premature node expansion, which in turn eliminates the price advantage of commodity servers. HCI 2.0 invests more in the per-node hardware making sure it can fully utilize the capabilities of modern high-performance CPU, internal connectivity and storage media.
The Hardware Matters for HCI 2.0
To prevent the node sprawl associated with HCI 1.0, HCI 2.0 utilizes the latest Intel Processors and PCIe attached NVMe flash devices to deliver maximum virtual machine density while supporting storage-IO intensive workloads. Beyond just supporting the latest CPU, HCI 2.0 delivers more storage per socket or node and since storage is the most common resource that forces expansion, the more capacity per node the less frequently new nodes need to be added. Where HCI 1.0 encouraged node sprawl by using generic 1U servers, HCI 2.0 maximizes per node capabilities by maximizing capacity per node which also means more devices per node which further reduces latency as well as reducing wear. As a result, HCI 2.0 has the computing power to support hundreds of virtual machines per node.
The computing power of HCI 2.0 is useless if it does not have storage performance that can handle the IO requirements of hundreds of virtual machines per node. HCI 2.0 leverages PCIe attached NVMe flash drives, which ensures that these drives reach their maximum performance. HCI 2.0 also ensures that each VM’s data IO path is local to that node, meaning no network induced latency. Nodes also, instead of only supporting a dozen or so drives per node, now support as many as 72 drives per node, which means that each node now has the processing power, the storage IO capabilities, and the storage capacity to support hundreds of virtual machines and a wide variety of mixed, storage-intensive workloads.
For protection and VM mobility, an HCI 2.0 solution replicates data to other nodes in the cluster. Replication allows for a VM’s data to be intact on the node on which the VM is running, which improves storage IO performance. It also significantly reduces network IO, which simplifies cluster configuration and management. Some HCI 2.0 solutions also leverage software-defined networking attributes so that the addition of nodes to the cluster and other network operation needs are automated.
The Proprietary Workaround
The absence of a more hardware-centric approach to HCI is leading some vendors to create custom hypervisor solutions, which often still don’t address the raw performance needs of the data center but may lower licensing costs. The switching costs of moving from a mainstream hypervisor to a more customized one are significant. Not only do hypervisor operations need relearning, but the customer also has to re-purchase the various utilities that go along with the custom hypervisor, if they exist at all.
HCI 2.0’s more hardware-centric approach enables mainstream hypervisors, such as Hyper-V and Storage Spaces Direct, to deliver the performance required by storage intensive Tier 1 workloads as well as support for a wide variety of workload types. The customer can continue to leverage its knowledge of the hypervisor and the operating system, with which it is familiar, as well as the utilities it uses with it. Most importantly, the proprietary hypervisor approach still doesn’t address the core IO problem. Without the right approach to the hardware it can’t deliver the performance that storage intensive applications require.
In many ways, HCI 2.0 takes the exact opposite approach to scale that 1.0 solutions take. HCI 2.0 focuses on per node performance capabilities, and while it can scale-out, it attempts to slow the rate of scale-out. The more the architecture can “scale-in,” the better the performance each virtual machine experiences and the lower the overall costs of the hyper-converged architecture.
In our next blog we will discuss how HCI 2.0 is much simpler to expand and operate than HCI 1.0. We will also take a closer look at the total cost of ownership of HCI 1.0 vs. HCI 2.0.
In the meantime, sign up to join Storage Switzerland and Axellio Inc., for our on demand webinar “How to Put an End to Hyperconverged Silos.” In this webinar you’ll learn why current generation HCI solutions fall short and the essential requirements for HCI’s next generation.
- Learn Why HCI 1.0 Shortcomings Are Costing You Money and Adding Complexity
- Learn Why Hardware Matters in HCI Solutions
- Learn How HCI 2.0 is ideal for Tier 1 Workloads
Register now and receive a free copy of Storage Switzerland’s latest eBook, “What is HCI 2.0?”.