How Do You Slice a Network?

Network slicing www.apistraining.com 5G trainings for engineers In the world of telecommunications today, and particularly in the somewhat chaotic domain of 5G, the term “Network slicing” is thrown around a lot. It seems important. It’s mentioned by all the big standardizing bodies, and often named as a cornerstone of future telecoms. It’s also mentioned around a lot of offices and water coolers around the world, and perhaps you’re just like I was and don’t really have the nerve to say, “So what is this network slicing everyone keeps talking about?”.

So I’ll tell you.

By the way: I’m Martin, and I’ve been a telecom technical instructor since the last millennium. Or counting in a way that may be more relevant to you: since 2G. For a long time that meant teaching special telecom things that only existed in the telecom world. With the advent of Voice-over-IP and SIP, my telecom world started crashing with datacom, and that crash only continued when virtualization entered stage. Suddenly telecom, datacom and general computing was all mixed up!

So, since a couple of years, we’re in this chaotic, mixed-up world.
How about I actually go on with the story of network slicing, then?

Network Slicing

First of all: network slicing is not new. Much like many (some might even say all) of the constituent parts of the 5G revolution, it has been around before, and the new thing is sticking all these parts together instead of keeping them apart. Perhaps it was really chaos before, when we had so many different technologies that didn’t interact, and now when everything starts to get connected, it’s actually bringing order to the world? Let’s stick with that story for a while, instead. Sounds more positive.

Technical Part

Technically, network slicing is very comparable to cloud technology. And to not make things even vaguer, by cloud technology I mean hardware resources that can be used by many customers, but are not dedicated to customers over time. Example: Dropbox. You get access to a hard drive somewhere, that is physically shared with others, but you never see them. Or a server you rent from Amazon to run your web page on – the physical server probably has other customers as well, but you never see them. Or a fiber optic cable connecting Europe with North America. It’s extremely likely you’re not alone on this cable, but again: you never see the others. In all these examples you can easily change the service (increase or decrease resources) without having to care about any physical reality. Buy more storage space – buy another virtual web server – add more bandwidth. All at the swipe of a credit card!

One way of phrasing this is to separate physical networks from logical networks. Even though you share all the physical aspects of this world (hard drives, servers, cables) with others, you have your own logical hard drive, your own logical server, and your own logical transatlantic connection. Logical things that don’t interfere with other customers’ logical things. Logical is a word being used a lot here, so you have to accept that, but it’s also a word that sounds more vague and nebulous than it is. Like “virtual servers”, “ghost computer” or “imaginary number” the words themselves almost makes it worse, don’t they? In our case a “logical” resource given to you just means that during a particular span of time you are given access to a particular portion of a physical resource. That portion is dedicated to you during that time, but may be bigger, smaller or not yours at all at a later time.

So what happens if you connect all your logical things? You connect your logical hard drive (a portion of one or several physical hard drives) to your logical server (software using a portion of the CPU and RAM of a physical server), and then connect that server to your logical network connection (a portion of the bandwidth available in a physical network connection). And then perhaps add another logical server on the other side of that logical connection. Well, then you have your own network to play with, that others can’t see. Then again, others could (and will!) do the same thing and connect their logical servers/hard drives/cables into a network that you can’t see.

Remember that this is all running on the same physical hardware. You and those mysterious “others” all share the same real-world infrastructure. It’s almost as if we had “sliced” that real world of hardware into logical slices containing servers/hard drives/cables and given one slice to you and one slice to me. Yes, let’s do that, let’s call them slices. Because if we do, we have performed “Network slicing”. Network slicing is the act of having one set of hardware – one physical network – serve several customers or uses, each slice being a “complete” network, but also not being aware of the other slices.

You can also see this as the next step up for logical things – remember those? Logical things were portions of hardware resources. Portions of hard drives, servers and cables. A logical network (or network slice) is then a portion of a physical network, including portions of the resources available (storage, CPU, network). And just like you could change your allotment of logical storage or logical bandwidth, so can your whole logical network (or network slice) change over time. More customers  – grow with a few extra nodes and more bandwidth. Early morning – shrink your resources, because everyone is sleeping. All at swipe of your metaphorical (or real!) credit card.

5G Use Cases

In 5G, network slicing is often used in the context of the different use cases. There are currently three well-known 5G use cases:

1. Super-fast Internet (eMBB or extreme Mobile BroadBand)
2. Super-good network connections (URLLC or Ultra-Reliable and Low-Latency Communication)
3. Internet-of-Things (mMTC or massive Machine-Type-Communication).

These use cases all present different networking challenges and the networks that provide them can reasonably be expected to look quite different. But instead of building three different networks, operators build one. A flexible one, though, that can be sliced up into slices perfectly matched for the use cases you need – and when you need them!

How to slice?
If you wonder exactly how you slice a network into pieces, the answer is either very short or very long. I’ll give you the short one: NFV and SDN. And if you want to learn the long answer, just come to us at Apis IP-Solutions and ask. We’d love to tell you!

Watch our YouTube Video about Cloud, SDN and NFV training

[youtube]https://youtu.be/6jIDRNTe-LI[/youtube]

Leave a Reply

Your email address will not be published.