31 October, 2022
Explaining SymplyPERIFERY: nodes and clusters
Welcome to Part 1 of our short series explaining SymplyPERIFERY.
Here at Symply we tend to talk about storage a lot. Which makes sense, since we build storage solutions. One of our largest to date is SymplyPERIFERY - a scalable S3 native appliance that’s getting a lot of attention. There’s plenty that goes into understanding exactly what PERIFERY is, what it does, and how it works. More, in fact, that would fit into a single video or a very dense PDF. So we’ve created this seven part video series to explain the core concepts. Welcome to episode one.
SymplyPERIFERY is a software-defined scale-out storage platform. There are two main parts you need to understand: the hardware (which is designed and built by us) and the software (which is made by a company called DataCore).
The names of these two parts can be a little confusing. Our hardware is called SymplyPERIFERY. DataCore’s software is also called Perifery. We’ve deliberately teamed up to use the same name so users understand that one comes with the other.
DataCore’s software powers our physical appliance - it’s shipped pre-installed to create a robust storage cluster for data protection, management, organization, and search at a massive scale. The key to SymplyPERIFERY is its parallel cluster technology. This isn’t an original idea. In fact, it’s taken from hundreds of millions of years of evolution: insects. Or, in fancy words, the emergent smart behaviour of swarms of nearly independent agents — in our case: storage nodes… and not insects. These nodes share a minimal set of sensors, parameters, and objectives that allow them to carry out their tasks with the greatest of efficiencies.
So SymplyPERIFERY as a complete solution is made up of nearly independent nodes that form a larger storage cluster. Those nodes, and that cluster, are all governed by DataCore’s Perifery software.
But why a swarm-like cluster? Well… some geometrically inspired cluster configurations, like rings, are good in theory but in reality a cluster is a harsh and less than deterministic environment. Only very simple rules and concepts survive at key defining moments and that can make or break a system. With SymplyPERIFERY architecture if the data can be reached, it can be read - even in the most extreme networking situations. And the independent nature of our storage nodes means they can’t drag each other down — or, critically, drag down the whole cluster. That means SymplyPERIFERY is essentially impervious to hardware failures, and deals with emergency situations in a sensible, autonomous fashion.
So let’s summarise:
SymplyPERIFERY is a collection of nearly-independent storage nodes connected together to form a cluster. Think of it like a beehive - dozens or hundreds of insects all operating independently, but simultaneously as a group. It becomes scalable because of that nearly independent behaviour. Just add another node (or another bee in our example) and the software will recognise it and adapt to it. DataCore’s Perifery software manages all those nodes and gives them rules to work to and with.
That’s the setup. Storage nodes connected together with a defining set of rules over the top. In our next video we’ll talk about how the drives inside the nodes are used and what networking choices you might have.