Written by Herb Villa, Senior Applications Engineer - Rittal North America LLC
No matter how unpredictable the world is, there is one thing you can count on: technology will continue to evolve. Everyone has heard of The Edge. But how do you get there?
Two decades ago, the introduction of Edge Computing changed the game, and installations started popping up in places far from traditional data spaces (under a cell tower, in a subway, on the factory floor, etc.)
We know we start in `The Cloud'. And we know The Edge is far removed from the enterprise data centre; way, way out there at the furthest limit of the network, right next to the processes, machines and sensors generating all that data. But is there something in between something that provides an intermediate stop from The Cloud to The Edge?
Edge computing efficiencies can be enhanced using a hub & spoke distribution model to enhance connections and efficiencies. It is a technique being used now, yet no one has established a proper name, nor defined what it actually is.
Some sources call this an Edge Data Centre,but that confuses two different deployments. By definition, edge deployments are placed where data centres (large, secure buildings) can't go and don't belong. To use them together in one name muddles the waters instead of clarifying them.
Spine is Rittal's way of explaining this functionality while rejecting misleading phrasing; and here's why it is appropriate.
Think of an airline. Its largest hub may be in Houston, yet there are many other hubs around the country: San Francisco, Los Angeles, Chicago, Denver, New York, and Washington DC. There is no single, main headquarters for all of its operations, yet they are all connected into a network. Now, look at the many small airports in remote locations as Edge locations. Between the big hubs and the remote locations are what Rittal would call the spine airports, the connectors that keep the network flowing in the hub & spoke plan (see Innovative 3 Layer Topology below).
In the graphic above, the Edge deployment is a standalone deployment: one or two footprints with climate control, power, security, cable management, etc. just like it is in the Core deployment. From hyper-local (EDGE) dge to hyper-scale (large data centres, or Core), the biggest difference is the scale, not the contents.
The Spine also has the same contents, but it's in the middle, receiving input and data from multiple Edge deployments and aggregating them. The spine contains network switches to filter and forward, and it also connects with other spines without having to go through the core.
Similarly (see below), the spine can act as a data centre while connecting Edge deployments to the cloud.
Edge Standalone Diagram
Unlike an Edge deployment (as explained above: located anywhere BUT in a traditional data centre), a Spine deployment can be located in a dedicated data centre space, or it could be in a closet. Maybe a container located somewhere on the grounds of a factory, medical centre or school campus. Obviously a 20-enclosure spine deployment won't fit in an office closet down the hall, but one or two easily can. Three or four enclosures in an unused office Sure. Ten in a repurposed classroom Okay. It all depends on need and what is available.
How many footprints are in the typical spine deployment?
Instead of hundreds of enclosures in a large data centre, the spine is
around 10-20 footprints (or whatever number is enough to handle
the Edge deployments it is assigned to support).
Are you using a spine deployment and not know it? Maybe you just haven't called it Spine. If what you call an Edge Data Centre is located in a traditional data centre, it likely is not Edge. If it is located in a non-traditional environment (factory floor, etc.), it likely is Edge and could be Spine if it connects multiple Edge deployments.
With What is it? and Where is it? answered, let us dive into Why Spine exists? Simply put, it provides local, higher levels of support for Edge applications, increasing process capabilities at the local level and interconnecting without having to go through the core.
A spine deployment expands efficiency in numerous ways:
Understanding gained efficiencies, for instance, is as easy as thinking back to our airline analogy. Flying from New York to El Paso requires a changeover at the Houston hub. One stop isn't nonstop, but it's still a quick trip. If that route included flying from New York to Miami, then to Memphis, onto Houston, and then to El Paso, time triples, efficiency is gone, bandwidth is reduced, and resiliency is shot.
The hub-spoke distribution model comes alive when scaling from multiple Edge standalone platforms to a localized, concentration point in order to effectively process data.
Yes, spine adds a layer of IT management, which is critical to improving efficiency. Test this yourself. Google Edge Computing and see what pops up. Boom! Enjoy your 1,330,000,000 results in 0.62 seconds (no kidding; those are the real results.) That is the advantage of management at the software and application level. It does an incredible job at filtering, so the necessary results are shown in the most relevant order, according to Google, of course.
So moving forward, Edge Data Centre doesn't exist. The deployments that are called this not in a traditional data centre, yet large enough to hold a dozen footprints are actually spine deployments, providing efficiency benefits across the network.
Ready to learn more about Spine Edge functionality? Talk to the experts at Rittal.
You might also like: What are the main benefits of Edge Computing