Product Announcements

Understanding The Layers Of Hyper-Converged Infrastructure

We’re almost half way through the year, and it is shaping up to be a “tipping-point” year for hyper-convergence and HCI. Combine faster CPUs, lower cost flash (with exciting technologies on the horizon) and software innovation with the majority of data centers standardizing on server virtualization, now is the time to extend existing infrastructure investments with newer, modern software-defined solutions.

Three months ago, VMware introduced Virtual SAN 6.2 and gave our hyper-converged infrastructure (HCI) software stack a name: VMware Hyper-Converged Software (VMware HCS). Virtual SAN 6.2 introduced a major set of new features to help improve space efficiency and management (check out the What’s New in Virtual SAN 6.2 blog for those details). VMware HCS is the name used to refer to the software stack of Virtual SAN, vSphere and vCenter Server.

With all the various terms and names being used to refer to HCI and the components, I want to take a few minutes to help clarify the terms we use at VMware and break down our view of HCI.

Does Virtual SAN = HCI?

We sometimes use HCI, VMware HCS and even Virtual SAN in similar ways to refer to a solution where compute and storage functions are delivered from the hypervisor software on a common x86 platform (i.e. HCI). While all those terms are related to HCI, they refer to specific components or groups of components that make up a full hyper-converged infrastructure solution.

It’s important to understand that Virtual SAN on its own is not hyper-converged infrastructure. Virtual SAN is software-defined storage that is uniquely embedded directly in vSphere. Virtual SAN refers to the software that virtualizes the storage layer by abstracting and pooling together the direct attached storage devices (SSDs, HDDs, PCIe, etc…) into shared storage. It delivers all of the standard storage data services, from availability to snapshots to deduplication.

Because Virtual SAN is so tightly integrated with (and dependent on) vSphere, whenever you talk about running Virtual SAN, the understanding is that the compute virtualization piece from vSphere is there too.

Similarly, vSphere with Virtual SAN requires hardware to run it—as someone reminded me recently, software without hardware is about as useful as an ejection seat on a helicopter (think about that one for a sec if needed).

As the image to the right shows, HCI refers to the overall solution that includes two major components: hyper-converged software and industry-standard hardware. Without both of those pieces, you do not have HCI. From VMware, our software stack is VMware HCS, but an HCI software stack can look very different across the various vendors in the market.

VMware has a unique advantage in that VMware HCS is a tightly integrated software stack embedded in the kernel and is the only vendor that provides such level of integration. This architectural advantage delivers a number of benefits including: performance, simplicity, reliability and efficiency.

Do all HCI solutions look the same?

While all HCI solutions generally follow this blueprint of having a software stack built on a hypervisor that runs on industry-standard hardware, in the end they can look very different and can have varying degrees of integration.

HCI solutions start with server virtualization (some hypervisor, which is more times than not vSphere) and then add in software-defined storage capabilities. Software-defined storage can be delivered tightly integrated with the hypervisor, like Virtual SAN, or bolted on as a virtual storage appliance (separate VM on each server). That software stack is then loaded onto an x86 platform, with industry-standard components (drives/SSDs, controllers, memory, …).

Some vendors package that together into a turnkey appliance that can be bought as a single SKU, making those HCI layers less transparent and the deployment easier. One example of that type of HCI solution includes VCE VxRail HCI Appliance (which we’ve done with EMC) and is built on the full VMware HCS stack.

VMware HCS also offers you the ability to customize your hardware platform. You can choose from over 100 pre-certified x86 platforms from all of the major server vendors. We call these hardware options our Virtual SAN Ready Nodes.

An advantage to the Ready Node approach is that you can choose to deploy hardware that you already know. Equally important, but often overlooked, is that the relationships that you have with a partner or vendor, the procurement process you have in place and the support agreements with your preferred server vendor can all be leveraged. No need to create new support and procurement silos. No need to learn a new hardware platform including how to manage, install and configure it.

Get Hands-On with VMware HCS Today

If you want to see these components in action, then you can test-drive Virtual SAN today for free with our online hands-on labs. It is lab HOL-SDC-1608.

This post originally appeared on the Storage & Availability blog Virtual Blocks and was written by Michael Haag. Michael is a Product Line Marketing Manager in the Storage and Availability Business Unit at VMware, specializing in software-defined storage (SDS), hyper-converged infrastructure (HCI) and traditional SAN storage with virtualization.

Related Articles