Primer Series: Cisco HyperFlex
In This Article
Have you ever sat in a meeting and the conversation turns to a technology with which you are unfamiliar? Suddenly, a bunch of acronyms are being thrown around. You have no idea what they mean, while everyone else is nodding their heads and seems to know exactly what is being discussed.
We've all been there, and to help our valued customers, we've decided to write a series of 'primer articles' to give the reader basic information on various products and technologies. This article is going to cover the basics of Cisco's hyper-converged infrastructure (HCI) technology -- HyperFlex.
Before we dive into the technology, we highly recommend you take a look at the Hyper-Converged Infrastructure Primer article, as this article requires some basic understanding of HCI concepts.
Cisco hasn't historically been known to provide many technologies involving storage. Besides thinking of your typical networking equipment when someone says Cisco, in the data center space, we tend to default to Cisco's Unified Computing System (Cisco UCS servers) or converged infrastructure (CI) solutions like FlexPod or FlashStack. In the CI world, Cisco is used as the compute layer but not the primary storage provider. HyperFlex changes this narrative.
The HyperFlex software defined storage (SDS) layer, formally known as SpringPath, has been around for quite some time. Cisco officially acquired SpringPath in September 2017 after successful testing and increased market interest. If you hear someone interchange HyperFlex and SpringPath, they are essentially the same, as you can't have one without the other.
- SpringPath: The former name for the software that manages the local disks.
- HyperFlex: The entire solution including the x86 UCS servers, Fabric Interconnects (FIs), a hypervisor and of course, the SDS layer where all the magic happens.
- HX: The abbreviated name for HyperFlex. You'll hear HX / HyperFlex being interchanged.
A quick side note, Cisco does have a HyperFlex offering that does not require FIs. This is called HyperFlex Edge and is typically used for remote office/branch office (ROBO) scenarios. We won't dig into the offering as part of this article, but you can find more information from Cisco.
We also want to point out a simple, yet very important aspect of HyperFlex. We've had countless discussions on the HCI topic and every now and then, there's confusion with the VMware vSAN product and how it ties in to HyperFlex. Let us be perfectly clear:
HyperFlex does NOT require vSAN.
As a matter of fact, you cannot have two SDS layers operating on the same servers. It is not supported and honestly just won't work. They are competing technologies. That said, HyperFlex does require a hypervisor in the form of VMware vSphere or Microsoft Hyper-V.
Why do we point this out? We've seen instances of HyperFlex being quoted out with VMware vSAN licensing, needlessly increasing the cost of the solution. The above paragraph also applies to any other SDS technology that can run on top of Cisco UCS hardware. HyperFlex is your storage provider and must take full control of the physical disks.
For the most part, HyperFlex follows the basic rules of the HCI concept with a few differences that we'll point out below but before we do, we need to explain a few terms.
- Cisco UCS C-Series Servers: Cisco has several server models, however, HyperFlex is currently only supported to run on top of C-Series servers, which are rackmount servers that come in 2 form factors: 1 rack unit (C220) or 2 rack units (C240). Find more information.
- Fabric Interconnects and UCS Manager: Fabric interconnects (FIs) are essentially switches with intelligent software called UCS Manager running on top of them. These act as your storage switches which is a major differentiator for HyperFlex. Find more information.
- CVM: In the HCI Primer article, we covered what the CVM is: Controller Virtual Machine. This virtual machine is the SDS layer for HyperFlex. This may also be referred to as the SpringPath software.
- HX Connect: HX Connect, or HyperFlex Connect, is the web interface that has been developed for on-premises HyperFlex management. Most of these functions can also be found in the HyperFlex plug-in which is installed on your hypervisor management interface during deployment. This includes things such as monitoring, alerting, storage configuration and native replication.
- Intersight: Intersight is a cloud based Software as a Service (SaaS) offering from Cisco. It allows customers to deploy, monitor and upgrade HyperFlex clusters (among other things). We guarantee you'll hear this term during a presentation, which is why we also wrote an Intersight Primer article.
The image above brings the terminology together in a visual fashion. Some of the more technical terms are found below.
- IOVisor: The IOVisor can be thought of as the storage network traffic cop. It will direct where writes and reads should go and come from. The IOVisor is a differentiator for Cisco as it allows for a more dynamic distributed storage architecture across nodes in a cluster.
- HX Acceleration Engine: The acceleration engine is a fancy Field Programmable Gate Array (FPGA) offload card which currently gives HyperFlex the ability to have additional compression savings. It is not required as part of the overall solution but should be considered depending on the workload.
- NVMe/SSD: These make up the usable storage layer for virtual machines. NVMe stands for Non-Volatile Memory express, and SSD is solid state drives.
- HX Profiler: HX Profiler is a free tool offered by Cisco to monitor workloads that are being evaluated to run on top of HyperFlex. It is not required to run HX Profiler but highly recommended to ensure proper sizing.