The Continuing Evolution of Hyperconverged Infrastructure

It doesn’t seem all that long ago that the market was introduced to what we now consider “original” hyperconverged infrastructure technology.  In that initial wave of technology, we saw appliances combining servers and storage hit the shelves, firing a shot over the bow of storage providers that the status quo wouldn’t remain the status quo much longer.  Since then, we’ve seen player after player hit the market with new hyperconverged solutions and we’ve even seen a few companies fail to hit the mark.
Of course, every hyperconverged infrastructure vendor on the market today has some differentiating feature, but at the core, whether delivered as a software-only solution or as a hardware appliance, most solutions have focused on solving the storage problem by combining it with servers and making the entire solution easier to manage and easier to scale while, hopefully, being less expensive than what came before.
Today’s hyperconverged infrastructure market is a far different place.  In this article, I’ll focus on four facets of the market that are pretty interesting:

• Evolution of the network in hyperconverged solutions
• The role of the Cloud
• Bringing certainty to an uncertain world
• Some SMB love

Bring Out Your Networks!

One resource has been conspicuously left out of the hyperconvergence party: the network.  Given the heavy reliance on the network in a data center full of interconnected appliances, the network can become a sticking point.  Most hyperconverged infrastructure solutions require linking the appliances together with a 10Gb fabric, although some will work with 1Gb connectivity.  Regardless, the network – that linking fabric – has generally been outside the control of the hyperconverged infrastructure vendor.
The lack of network control means that, while applications on a node may be able to be subjected to strict SLAs and performance requirements, no such had guarantee could be made between nodes or between a cluster and your users.
To combat this, we are beginning to see solutions hit the market that assimilate networking into the hyperconvergence equation in new and different ways.  My colleague James Green previously discussed one such approach from Cloudistics, which provides a hyperconvergence-on-steroids solution they refer to as superconverged infrastructure.  The Cloudistics solution ships with switches that run the company’s proprietary code and that integrate deeply with the compute/storage infrastructure to provide fast, efficient, predictable connectivity.  It’s truly an all-in-one core data center solution.
Another company doing interesting things in the world of networking and hyperconvergence is Stratoscale.  A software-only solution, Stratoscale brings an abstracted overlay network that brings networking features natively into the hyperconverged environment.  Capabilities include virtual switching, routing, NAT, security groups and other L3/4 services, integrating with software defined networking solutions for advanced networking use cases like service chaining.
Finally, just today, HyperGrid – a mashup of Gridstore and DCHQ – was launched.  HyperGrid includes a pair of top of rack switches in their solution, obviating the need for customers to procure their own and ensuring full compatibility between all components in the hyperconverged infrastructure solution.  Simply plug in red and green cables and go on your way.

Hybrid Cloud

The cloud is something hanging over every IT pro and CIO’s head, both literally and figuratively.  For some, the public cloud is the target for all new workloads.  For others, they are just seeking the benefits of cloud, but still want to maintain their own environment.  I recently wrote a post describing the kinds of benefits that people are looking for in their data center environments.
So, some are looking for public cloud, some are looking for private cloud, but, in reality, most will move forward with some combination of the two.  For these hybrid cloud adopters, flexibility in workload deployment is key.  For some, the ability to shift workloads from on-premises to a cloud provider is key as well.
Hyperconverged infrastructure systems, which vendors are pushing to create a private cloud foundation, have long been able to integrate with public cloud providers to create a hybrid cloud computing environment.  For example, SimpliVity’s software can run in a public cloud environment and integrate with on-premises appliances to enable such disaster recovery scenarios and other functionality.
The aforementioned HyperGrid is another company that is making the on-ramp to hybrid cloud far easier than it used to be.  The company’s HyperForm software solution working with the HyperGrid hardware appliance and also supports integration with 18 cloud providers.  On the private side of the equation, HyperForm supports Hyper-V, vSphere, OpenStack, and more.

Bringing Certainty to an Uncertain World

When it comes to data center infrastructure, there are no guarantees.  Or are there?  In recent years, as upstarts have sought to topple the biggest players in the industry, they’ve had to grapple with the fact that they have little to no track record.  Further, when it comes to hyperconverged infrastructure, there are often claims that, at first, seem seriously silly.  However, time has proven out many of the claims made by vendors.  In fact, we’ve seen some vendors in the space go so far as to guarantee certain outcomes based on the kinds of workloads that are in operation at the time of purchase.
For example, one of SimpliVity’s guarantee clauses says this: “If you use SimpliVity hyperconverged infrastructure and its built-in VM-centric backup capability as outlined, you will save 90% capacity, across storage and backup combined, relative to comparable traditional solutions.”  That’s a 10:1 reduction ratio.  In the interest of full disclosure, this guarantee does have some requirements, including using SimpliVIty to handle your backups, but, regardless, guaranteeing a 10:1 data reduction during the warranty period is pretty incredible.
Likewise, HyperGrid provides what they call a 5×5 Guarantee.  Under this guarantee, the company promises you’ll see cost reduction by at least 50% and that you’ll receive a number of other performance benefits.  If they miss during this 30 day period, they’ll take back the appliances.
These kinds of guarantees are far more than we typically see from legacy vendors and they provide customers with peace of mind for what can be significant investments.

Some SMB Love

The last evolution that I want to discuss revolves around the potential for the SMB to adopt hyperconverged infrastructure.  With one notable exception, most of the hyperconverged infrastructure solutions on the market were, until recently, focus at the midmarket and higher.  While SMB’s weren’t necessarily blocked from such purchases, the solution acquisition cost was often prohibitive.
That notable exception is Scale Computing, who provides solutions that are 100% focused on the SMB and small midmarket.  That’s been their space from the beginning and they are continuing to expand their capabilities through the introduction of modules sporting flash drives.  Scale allows customers to go all-in on servers, storage, hypervisor, and management with pricing that is incredibly SMB-friendly.
Other hyperconverged infrastructure vendors are also making their way into the SMB space, but they’re coming down-market from their enterprise solutions to do so.  Nutanix, for example, released Nutanix Xpress earlier this year as a way to broaden their customer base.

Summary

As we continue the hyperconvergence journey, I fully expect that we’ll continue to see continued expansion of the solutions that are on the market.  We’ll see more of them add public cloud integration capabilities; we’ll see more – and deeper – assimilation of the network resource; we’ll see more backing their products with guarantees; and, we’ll see more push down into the SMB in order to capture that vast market.