Over the past ten years, hyperconvergence transformed IT infrastructure by uniting server, storage, and hypervisors. This led to significant reduction in costs and complexity for many enterprise companies, but due to limitations in the ability to reduce the size of nodes, many small and medium-sized businesses were unable to achieve the same efficiencies. In this blog, we explain how HCI has evolved over the past ten years, and why Scale Computing built the first system that helps SMBs scale their IT infrastructure down, in order to scale their businesses up.
Two Types of Hyperconverged Solutions
There are two types of ways in which hyperconverged solutions handle storage management. The first one is to create a virtual machine dedicated to managing all of the storage resources on a local node. This is called a virtual storage appliance or VSA. This type of solution is generally more inefficient and requires more overhead because the virtual machine is in the data path for all I/O operations that take place on the host and betweens hosts in the cluster.
The second option is to use a hypervisor operating system to handle storage operations. Rather than a controller virtual machine, this module is the key player in the data path to handle storage operations. This is how we architect our solutions at Scale Computing due to improved performance and integration with the rest of the I/O stack.
Evolution of Storage Locations
Most organizations start out very small, often in a single location. Managing IT infrastructure in one location, with IT on-site, is fairly easy. However, as companies grow and require new locations to manage customer service, distribution, additional franchises, etc. it traditionally required IT to build complex systems where on-premises hardware was housed at each building, but managed in the central office. Predictably, this created issues with lack of support since it would be costly to have dedicated support staff at satellite offices; IT organizations struggled to find space and cooling systems for servers; and increased costs because this approach required multiple, expensive multi-tier IT deployments.
Supplying Affordable, Dependable IT Infrastructure Solutions at the Edge
The collection of use cases and computing needs in satellite locations (like security and point of sale systems) are collectively called the edge, and edge computing is what powers these use functions. Often, edge computing requirements are lesser than what is needed at the primary data center, so nodes can be smaller yet they still need to be highly available, all while being cost-effective and simple to use. Historically, finding this type of a solution for small and medium-sized businesses was difficult because hyperconverged solutions were generally built for enterprise companies where cost is less of a concern. In addition, most hyperconverged solutions are incapable of making nodes smaller, so storage becomes a resource sinkhole.
Scaling Down So Businesses Can Scale Up
To solve these problems, we launched our HE100 and HE500 series of appliances that consumes a fraction of the processing overhead of other solutions. The Scale Computing Reliable Independent Block Engine (SCRIBE) storage engine is purpose-built to provide highly available, scalable compute and storage services at the core and at the edge with as little overhead as possible. And, the reduced complexity of our solution means that even a store manager or IT generalist can perform maintenance on the cluster, including replacing failed drives, rebooting virtual machines, and rebooting nodes.
Avoiding Hardware Vendor Lock-In With Scale Computing
Another advantage of our hyperconverged solution is that you are not locked into a specific vendor’s hardware as you scale. The hardware you start with is not what you have to evolve with, and that the mixing and matching avoids large "forklift" upgrades in the future. For edge computing or distributed enterprise, you may start small with Intel NUC and progress to fully racked server clusters (or vice versa). In fact, our customers often mix hardware to match use cases (for example, Intel NUC at edge sites with other brands at core data center), or are mixing Scale Computing on new appliances with Scale Computing software on older gear, in which case the hardware doesn't even match within a single cluster. All of this means that you can build the most reliable, flexible, and affordable infrastructure for your business.
To learn more about our hyperconverged infrastructure solutions, check out this white paper: Going Small to Get Big: Why SMB and the Edge Have Emerged as Pivotal Use Cases for Hyperconverged Infrastructure.