MEC – your guide into 5G

One problem with summer holidays in our industry (or is it a benefit?) is that one tends to let certain things slide and to enjoy more time away from work – whether it be on a formal vacation or just by working a little less than our usual “40 hours” – a very loooong 40 hours – per week. I am certainly guilty of that this summer – and one of the things I am guilty of is not highlighting some really important output from the ETSI MEC ISG. But… as they say… better late than never. So here it goes, but let’s start with background and get to the cool things ETSI MEC produced as we go.

We’ve all heard that “MEC is a 5G technology” although what that means is not exactly clear. In fact, in my very first blog posting, I highlighted that this can lead to some of my (least) favorite “MEC myths”. Here those myths are, re-stated:

  • MEC is a 5G technology, so until I roll-out a 5G network I don’t need to worry about it
  • ETSI MEC will be made irrelevant as soon as 3GPP defines its AF/NEF
  • MEC is only needed before 5G, at which point CUPS (meaning the UPF) replaces it

Side note: yes, the first and third statements are in fact mutually contradicting. But these are myths, they don’t need to be mutually consistent.

Clearly, I disagree with all of these statements, but what is the truth “according to Alex”?

Well, first let’s start with a statement which I think is rather obvious, although some may consider it bold – 5G is all about Applications. Everything else – NR, MEC, CUPS, Network Slicing – are just tools we know we will need to run these applications at scale, globally, on the same network, at the same time and in an economically feasible fashion. If 5G is to succeed, we need to be able to do all five of these. However, as has been aptly written, 5G applications, unlike the target applications for the previous “Gs”, are highly heterogeneous with often contradictory requirements. The NGMN white paper, although a few years old now, remains an excellent source for understanding this in more depth. And so, in order to make 5G work, we need to do a lot of things very differently.

This should – and does – worry operators. The investment required to roll this out is huge. At the same time because of the disparate number of use cases, the business cases are often unclear. Adding to this is the experience of 4G. The huge revenues on which promise the 4G investments were based did materialize – but largely for the over-the-top web player who made little of the investment, not the wireless operators who bore – and still bear – the brunt of the costs. And so, 5G creates a serious business conundrum.

A reasonable way to address this would be to ask, well, can I start with what I already have. Specifically, is 5G all about application, can I roll out a 5G application on my existing 4G network? At least as a pilot?
Significantly, the answer is YES for almost every single 5G application – the only exceptions are those which truly require application-level latencies below 10 mSec. For some applications you need to do nothing at all with your 4G network. For many of them, you just need to create a landing zone for the application within the 4G network, so that it can place certain components close to the user. That is why you need MEC!

And so, MEC is a technology which enables you to roll out 5G applications on your existing 4G networks. Clearly, if the initial roll-out is successful and you need to scale it, you will need the various technologies that will turn your network into a 5G network – and importantly, 3GPP is specifying 5G to enable incremental evolution from 4G to 5G. However, you will be upgrading to 5G in response to a real business need, with a fairly clear understanding of the business aspects of the application. While risk still remains, it is substantially reduced. And so, MEC is a critical technology on the evolution to 5G and the details of how it serves that role can be found in a white paper we published some time back. 
1. But what about 5G itself? 2. Is MEC still needed in 5G? 3. And if it is, will the work of ETSI MEC ISG be superseded by 3GPP? The answers are (in order):

  1. 5G is awesome
  2. YES
  3. NO

We provide a detailed answer in the recently published white paper - I should have highlighted this one in a blog post a month ago as soon as it was published, but summer intervened. But let me give you a summary. When people say "MEC is only needed before 5G, at which point CUPS (meaning the UPF) replaces it” what they really mean is something like “local breakouts or mini-ePCs are only needed before 5G, at which point CUPS (meaning the UPF) replaces it.” The latter statement is true, but it is not a statement about MEC. Remember MEC stands for Multi-access Edge Computing – with “Computing” being key. It is much more about the Cloud at the edge – how to manage it, how to run applications on it, what services such a platform can offer, etc. Getting the traffic to and from it is, of course, essential. And while 4G requires extra-standard solutions, 5G standardizes this – so that part will change. But it is just one part of a much more complex system. The rest of the system remains unchanged – and is, in fact, agnostic to what “G” you are on. As for whether 3GPP will supersede ETSI MEC – this is just another instance of that “everyone is competing” myth I wrote about last time. We all have our job to do and what 3GPP and ETSI MEC do is by and large complementary.

Let me leave it at that and invite you to view the two white papers for details. Enjoy your summer!

Migration made easy
Achieving the lowest latency for delay-sensitive t...