Reduced complexity and smaller data center footprints are a couple of the benefits of hyperconverged infrastructure (HCI). So it makes sense that systems, which combine core storage and compute functionality into a single, highly virtualized platform, would play well in remote and branch offices (ROBO) and network-edge locations.
This week Pivot3 launched a hyperconverged infrastructure product specifically designed for ROBO and edge computing.
And don’t forget Riverbed SteelFusion, said Mike Leone, a senior analyst at Enterprise Strategy Group. “For some reason it doesn’t land on many HCI lists, but it should,” he wrote in an email. “They integrate their WAN optimization technology into their HCI solution, meaning organizations can consolidate yet another component at the edge. And with the network being so important at the edge, that’s a big differentiator for them.”
HCI can reduce capex and simplify both hardware and software management at remote offices, said IDC analyst Eric Sheppard.
“HCI is designed to be up and running quickly,” Sheppard said. “You’re coalescing all of the storage elements down to a cluster of software-defined, industry-standard servers. So rather than having an edge environment where you have disparate hardware that needs to be managed, you end up with a highly virtualized and software-defined solution where all of the management is super simplified through automation or centralized management.”
Like its enterprise data center counterpart, Acuity X3 uses NVMe flash architecture, and it includes Pivot3’s policy-based Quality of Service (QoS) engine.
Using super-fast NVMe flash in the hyperconverged nodes gives the platform its low-latency and high-performance features, said Mike Koponen, senior director, product and solutions marketing at Pivot3.
“What makes our use of NVMe flash different is that we have the ability to make really efficient use of that NVMe flash through our QoS engine,” Koponen said.
The QoS engine automates workload prioritization. “When a customer puts a hyperconverged workload on the system, they apply one of five performance policies to that workload,” he explained. A one means the workload is mission critical and requires the highest level of performance. A five means it’s a low-priority workload.
“Let’s say there’s competition for resources,” Koponen said. “With our advanced quality of service engine, customers can be confident that their less-important applications aren’t going to trash their more important applications at the least opportune time.”
ESG’s Leone said Pivot3’s QoS engine helps companies reduce their physical IT footprint while handling multiple workloads without sacrificing performance. “Pivot3’s QoS is somewhat of a gold standard,” he said. “It’s a true differentiator for them in the space right now.”
VMware’s remote office HCI product is also a smaller scale of its vSAN storage-based HCI. It can run in as few as two physical servers. It also includes VMware Pulse IoT Center for management and security of all edge systems/gateways and connected devices such as sensors.
“As more data is generated at the edge, customers don’t want to send all of that data up to the cloud just to send it back again as they are using it to make decisions,” said VMware’s Mimi Spier, VP of Internet of Things (IoT), at a Mobile World Congress press conference last month. “More compute will be required at the edge so we are taking the next step and leveraging our hyperconverged technology and building up end-to-end solutions with our partners to deliver micro data centers at the edge.”
These devices are ideal for branch offices, retail stores, banks, and factories, Spier said.
Edge HCI Considerations
But before deploying HCI systems in remote and branch office locations, companies should take cost into consideration, said John McArthur, an analyst at Gartner.
“They’d like to have HCI from an ease of management standpoint, but if that ROBO needs high availability, or would benefit from high availability, many of these high-availability configurations of these hyperconverged solutions are frankly just too expensive for these remote sites,” he said.
Smaller versions of data center hyperconverged infrastructure can still cost upwards of $50,000 to $80,000. “And that’s just too much,” McArthur said. “If they can get down to a $20,000-price point and if they can still hit performance requirements, that will fit.”
Before investing in HCI for edge computing, companies should first determine which workloads they need to run at remote sites.
“Run those applications in a test environment under a variety of failure conditions to make sure the performance and the recoverability — to get back to a protected state — are both acceptable,” he said. “Is hyperconverged infrastructure a good fit for a remote office? The answer is: it depends.”