Making sense of scale with hyperconverged infrastructure systems can be challenging given vendor rhetoric. We know that hyperconverged infrastructure features a scale-out architecture that provides flexibility and extensibility. Hyperconverged units can be installed incrementally, adding to the shared resource pool managed as a single global system, to accommodate requirements in capacity and performance.
But what exactly is a “hyperconverged unit?” Most vendors use the term “node” to describe its building block. However, not all nodes are created equal.
To illustrate this point, let’s say that you’re planning a summer BBQ. You’ve invited 50 of your closest friends over for some grillin’ and swillin’. For drinks, you decide to purchase beer.
Since you want to ensure you have a sufficient supply of drinks for your guests, you consider the differences between cases of beer vs. a keg of beer. A case of beer (24 12-ounce bottles) delivers 2.25 gallons, so you estimate that you need to have about seven cases (168 12-oz bottle) of beer on hand. An alternative is to get a keg of beer (15.5 gallons or 1,984 oz of beer—the equivalent of about 165 12-oz bottles).
So which is better? Both individual bottles and a keg will satisfy your requirements.
Individual bottles may present more of a challenge with respect to keeping them cold. You’ll require more cooler space and ice. There’s also more overhead with individual bottles: unpacking the cases and putting the beer on ice, having enough bottle openers on hand, and collecting and returning empties after the party. And will you really have visibility into resources available throughout the night?
A keg of beer requires one setup for cooling, and won’t require the same amount of overhead. Once the keg is tapped, guests can pour their own into party cups (but cup cost and cup waste are two considerations).
Aside from that, you should have enough brew for partygoers. It’s just a situation of managing the number units. So is seven better than one because it’s a bigger number?
This is a common scenario that SimpliVity sees when it comes to comparisons with other vendor’s solutions. SimpliVity has a number of hyperconverged infrastructure models available, supporting differing amounts of storage capacity, cores, and RAM. What they all have in common is the data efficiency feature in SimpliVity OmniStack Data Virtualization Platform. SimpliVity guarantees 10:1 data efficiency (although, on average, SimpliVity customers experience about 40:1 data efficiency). That means that SimpliVity offers at least 10x more effective capacity per node. With a higher upper limit per node for capacity, SimpliVity’s nodes are denser.
Going back to our party planning, a SimpliVity node is analogous to the keg, while competitors’ solutions are analogous to a case of beer. That means that the number of kegs needed to satisfy party requirements would be significantly less than alternative solutions sized like a case. Alternative solutions may boast that they can achieve greater scale because of the higher number of nodes. However, scale may be better measured “by the drink” … in this case, the number of VMs that can be hosted per cluster of nodes.