Edge computing, remote offices, branch offices (ROBO), small data centers, and IoT use cases are on the rise. Most research predicts that the amount of data stored outside the data center will be 40X more than what is stored inside the data center. The challenge for IT is equipping the edge with compute, and storage with a solution that meets the needs of the edge, while at the same time being cost-effective and easy to manage remotely.
Why We Need the Edge
Getting all employees to come to a single headquarters is a business model that no longer works. The company needs to be available where the talent is and not force relocation or long commutes. At the same time, the concept of having everyone work from home is inefficient and difficult for IT to manage. Another big driver for edge computing is small data centers where organizations need just enough compute and storage to run local applications (like retail outlets, healthcare facilities and oil rigs – to name a few). In most of these cases, the cloud is not an option either because cloud latency becomes a big issue, as does lack of productivity during a service interruption. The edge is the compromise by providing pods of local compute and storage that interconnects to other offices and the primary data center.
Edge Computing Challenges
The first challenge for IT when creating edge architecture is dealing with the sheer number of locations; sites can be measured in the hundreds if not thousands. Secondly, these sites are NOT data centers, they have little or, in most cases, no IT staffing. The on-premises equipment has to be simple, flexible and easily managed remotely. The second challenge is dealing with the cost issues associated with edge computing. Downsized data center solutions won’t work at the edge. They require too much hardware and require data center grade power and cooling. The final challenge is dealing with the various types of edge locations that an organization may require. There is no one-size-fits-all. IT needs flexible solutions to meet the diverse needs of each edge location.
Why Hyperconvergence Falls Short at the Edge
Hyperconverged Infrastructure (HCI), because it packages compute, storage and networking into a single unit, seems like the perfect fit for the edge. But, upon closer examination, it falls short. The first big problem is the typical HCI solution requires three nodes to get started. For many edge locations, this requirement leads to an excess of compute and storage. Many HCI vendors also require that customers buy all the hardware from them with the software. That hardware has, of course, premium pricing. For one location, that premium is justifiable. Across 100 or 1,000 locations, it becomes untenable.
Introducing StorMagic – Hyperconvergence for the Edge
StorMagic’s SvSAN is a hyperconverged solution designed for the edge. First, it is software only and runs on almost any hardware platform that the organization might have. Second, it starts as a two-node cluster. Each node mirrors its data, synchronously, to the other node. If one node fails, the other node can start virtual machines (VM), as needed. Finally, it virtualizes disk, flash and memory, all as storage components, and can automatically cache data between these storage types to optimize performance.
The value for IT looking to equip edge locations is they can start with their current hardware, load the SvSAN software on two hypervisor servers, providing both a cost-effective and flexible approach to edge computing needs.
Finally, SvSAN provides centralized management and deployment capabilities. The headquarters location can act as the “witness” for all sites, eliminating “split-brain” risk and requires very little networking bandwidth and processing horsepower. It also integrates with vCenter and SCOM, as well automation, through various tools and scripts.
SvSAN also addresses some of the concerns over traditional HCI solutions. For example, one of the big challenges for hyperconvergence is scaling specific resources. Typically adding a node means adding everything (compute, storage and networking). SvSAN can scale-up by just adding storage capacity or scale-out by adding compute-only nodes that access data from one of the converged nodes. The solution can even grow to a dedicated server SAN model where SvSAN runs on dedicated hardware.
Edge computing is on the rise. The problem is there are a limited number of solutions available to address its unique challenges. The Edge requires cost-effective, flexible solutions that provide central management. StorMagic SvSAN meets all these requirements, plus addressing a few weak points found in other HCI designs.