What Cisco's New Hyperconverged Infrastructure Is and Isn't Good At

Taking a critical look at HyperFlex, Cisco's first major hyperconvergence play

Bill Kleyman

April 5, 2016

7 Min Read
What Cisco's New Hyperconverged Infrastructure Is and Isn't Good At
HyperFlex, Cisco’s hyperconverged infrastructure solution (Photo: Cisco)


The set of factors companies have to consider when devising their data center strategy has changed drastically. They have to take into account things like cloud services, mobile workforce, delivery of services at the network edge on top of the traditional requirements to maintain uptime and anticipate growth in capacity requirements. This April, we focus on what it means to own an enterprise data center strategy in this day and age.

Today’s business is tightly coupled with data center capabilities. We’re seeing more users, more virtual workloads, and many more use cases for greater levels of compute density. There’s no slowdown in data growth in sight, and data center requirements will only continue to grow. Organizations have been working hard to improve data center economics with better underlying data center ecosystem technologies.

In comes hyperconverged infrastructure (HCI) -- a next-generation technology which tightly couples the virtual controller layer with its own operating mesh. Here’s something to remember: there are a number of similarities between HCI and converged infrastructure. However, the biggest difference comes in how these environments are managed. In HCI, the management layer – storage, for example – is controlled at the virtual layer. Specifically, HCI incorporates a virtual appliance which runs within the cluster. This virtual controller runs on each node within the cluster to ensure better failover capabilities, resiliency, and uptime. In these types of models, you begin to see how technologies around software-defined storage (SDS) impact converged infrastructure systems.

Businesses across all verticals are seeing benefits behind this tight integration with virtual technologies as well. HCI reduces complexity and fragmentation around having to manage resources sitting on heterogeneous systems; it can reduce data center footprints; and it can greatly reduce deployment risks with validated deployment architectures.

There’s clear demand in the market. Consider this: according to the latest Gartner Magic Quadrant for Integrated Systems report, “hyperconverged integrated systems will represent over 35 percent of total integrated system market revenue by 2019.” This makes it one of the fastest-growing and most valuable technology segments in the industry today.

A recent IDC report looked at the converged systems market in Q3 of 2015. The market generated 1,261 petabytes of new storage capacity shipments during the quarter, which was up 34.8 percent compared to the same period a year ago. Finally, the report gave a big stat showing the amount of growth in the converged market: hyperconverged sales grew 155.3% percent year over year during the third quarter of 2015, generating more than $278.8 million in sales. This amounted to 10.9 percent of the total market value.

And so, with a booming and demanding market, Cisco jumped into the HCI waters. In March, it announced a new line of hyperconverged infrastructure systems called HyperFlex.

Cisco HyperFlex Limitations

If you take a look at the overall Cisco platform, the biggest missing component was a powerful virtual controller helping manage storage resources. This is where Springpath comes into play. The result? HyperFlex, a system that combines software-defined computing in the form of Cisco Unified Computing System (UCS) servers, software-defined storage with the new Cisco HyperFlex HX Data Platform Software (Springpath), and software-defined networking with Cisco UCS fabric that integrates with Cisco Application Centric Infrastructure (ACI).

It’s a great start and a great technology for an organization that already has a massive user base with the UCS and networking ecosystem. The HyperFlex architecture integrates directly into existing Cisco management environments to allow for complete data center scale. Still, as many first releases go, there are some limitations:

  • Today, HX systems will only support VMware vSphere. However, other virtualization technologies like Hyper-V are in line for support in the future. For now, this type of architecture can make a lot of sense only if you’re a VMware-heavy shop.

  • As it stands, there is a limit of eight HX nodes per cluster. However, with the Hybrid option, you can add an additional four classic B200M4 Blades for more compute power. This gives the architecture the ability to run 12 servers in a hybrid configuration. There is a catch, however: in the Hybrid solution, the B200M4 local storage is not utilized by the Cisco HX system.

  • If you’re hoping to integrate an NVIDIA GRID card into the Cisco HX solution, you’re out of luck. Currently, graphics acceleration cards are not supported. However, there should be support for this architecture soon. There is a cool workaround however: in the Hybrid mode, where you can add B200M4 Blades, you can also integrate the NVIDIA GRID. You’ll still have to create a separate UCS domain, but at least you can integrate GPUs into the mix. With this in mind, you can still deploy powerful, multi-tenant environments around virtual application and even desktop delivery. You can create great economics as long as the use-case fits in.

  • Deduplication cannot be turned off and deduplication and compression are both in-line. This means that if you have apps or workloads that cannot work with deduplication or compression, deploying them on a Cisco HX architecture might be a challenge.

That said, there are numerous critical features that set this technology apart from any other HCI solution out there. One of those aspects revolves around the fact that HyperFlex comes with full network fabric integration. This type of integration allows administrators to create QoS policies and even manage vSwitch configurations that scale throughout the entire fabric interconnect architecture.

Furthermore, the Cisco HX system takes data from a different approach. Unlike Nutanix, which believes in “data locality first” before spreading data across, Cisco HX spreads data across all nodes at the same time. This is accomplished by first writing to the local SSD cache; from there, replicas are written to the remote SSD drives in parallel before the write is acknowledged. This, however, also brings up a use-case limitation: you won’t be able to deploy an app or workload that requires more SSD capacity than the system has available in cache.

Understanding Cisco HyperFlex Use Cases

Even though this is a 1.0 version, there are still a number of use cases where Cisco HyperFlex can be deployed:

  • VDI and Application Virtualization. Let’s be honest, this is a virtualization-ready ecosystem. This is especially the case if you’ve standardized on the VMware hypervisor. Management is done completely from the Cisco Data Platform Administration Plug-in for vCenter. This allows administrators to control their resources, monitor the environment, and provision workloads as needed. This direct coupling allows for easy VM management and direct integration with a software-defined data center (SDDC) ecosystem. Virtual applications live very well in these kinds of environments, as do virtual desktops. Still, keep in mind the graphics limitations and know which types of apps and workloads will work well on the HX system.

  • Remote office, small/medium/large data centers, branch offices. Instead of deploying large and bulky pieces of hardware, you can utilize Cisco HX for specific use cases or entire remote architectures. For remote office admins, they have fewer management points and can control their environments much more effectively. These types of systems aim to reduce overall data center footprints while still increasing density and simplifying management.

  • Test/Dev and non-critical workloads. There are some limitations which may prevent the deployment of critical applications. However, for sandboxing environments, dynamic resource control, testing, and workloads delivery, Cisco HX can make a lot of sense. If you plan on putting some mission-critical applications or workloads on the HX system, make sure to look at the underlying requirements and making appropriate adjustments.

As Cisco takes a big swing at a new market, it’ll be interesting to see the growth in the space and all the new use cases around HCI architectures. As for Cisco, how much will the HX System cost? According to CRN, the pricing for a three-node HyperFlex cluster starts at $59,000, including one year of 24x7x4 on-site support.

Even though Cisco is a bit late to the HCI market, they are leveraging a huge install base and very loyal customers. There’s no doubt there will be a number of use cases around Cisco HX solutions which will help change the way data centers are designed and deployed. Either way, the market is diving deeper into software-defined technologies as the virtual controller layer becomes even more intelligent.

About the Author(s)

Bill Kleyman

Bill Kleyman has more than 15 years of experience in enterprise technology. He also enjoys writing, blogging, and educating colleagues about tech. His published and referenced work can be found on Data Center Knowledge, AFCOM, ITPro Today, InformationWeek, NetworkComputing, TechTarget, DarkReading, Forbes, CBS Interactive, Slashdot, and more.

Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.

You May Also Like