28. Meryem Simsek, Bell Labs discusses What's Next for Edge Connectivity

Meryem Simsek, Head of Network Architecture Research at Bell Labs joins Elevate the Edge hosts Jo Peterson and Maribel Lopez for a discussion about What's Next for Edge Connectivity.

Transcript

[00:02] Maribel Lopez: Hello and welcome back to Elevate the Edge Podcast. As always, we are excited to be here today. I am Maribel Lopez and I'm joined here with my lovely cohost, Jo Peterson. Hey Jo.

[00:16] Jo Peterson: Hey Maribel.

[00:17] Maribel Lopez: And we are here today with Meryem Simsek. She is the head of Network Architecture Research at Nokia Bell Labs and I remember the various instantiations of Bell Labs and I'm really excited to have this discussion. So in her role, she leads a team in multiple worldwide locations to drive network architecture innovations across multiple network layers and across multiple network segments. Prior to her role at Bell Labs, she held leading research positions at VMware, Intel Labs and in academia. Welcome to the program, Meryem.

[00:55] Meryem Simsek: Thank you so much, Maribel. I am really excited to be part of this and looking forward to have a great conversation with you and Jo.

[01:03] Maribel Lopez: So there are many directions we could go in with Edge. One of the things that we have been talking about is edge connectivity and it's a large part of the computing equation. And obviously Nokia Bell Labs is known for research and innovation. And I was wondering how you and your team are working together on new network systems and device concepts as well as new solutions in maybe the application area or new network architecture innovations. It's a pretty large question, but basically I could maybe sum up the question as tell me what you guys are thinking about and working on researching and designing in the Edge related to connectivity. How's that?

[01:49] Meryem Simsek: Slightly sounds great. Yes. That's very concise. I will start with a very brief answer, but then expand it a little bit to provide more details in my team, which is a fairly new research lab that has been started after I joined in October. We are really pursuing a holistic approach, looking end to end, starting from the infrastructure, the transport network to the mobile and fixed networks all the way up to the applications and ask ourselves how can networks be designed for 2030 and beyond? That's the short answer, but looking into a very holistic approach. Of course, you are tackling multiple different areas within the network, but also multiple different technology trends, including edge. And what I would like to clarify here a little bit is you mentioned edge connectivity in your questions, which is really a part of the Edge compute computing equation. We talk quite a bit about edge everywhere. It's a fairly fuzzy word used for multiple different in multiple different contexts for multiple different purposes. And really what the edge computing is, it's nothing but a distributed architecture that processes data closer to the end users. Then on the other hand, we do have the network edge, which is basically the physical point where an enterprise network connects to a third party network services. And if we now talk about edge connectivity, then what we see is that with the increasing data acquisition devices deployed at these network edges or edge networks. It is actually really required to create a reliable wired or wireless communication environment that extends the power of edge computing to remote areas, distributed branches and edge data centers. And here edge connectivity in this context really addresses the performance and connectivity bottlenecks towards applications that are requiring very high degree of computing and wireless communication. That's what I would see under edge connectivity. Consumers in this context are for example enterprises that consume edge compute and meet very reliable and low latency connectivity to the distributed edge clouds which they depend on to host their very critical business cases and applications as well as workloads. And in this context there are a number of connectivity options that depend on the location, the compute provider, the performance requirements but also generally the availability of connectivity is that we can look here into wired connectivity and wireless. Connectivity and edge computing or computing itself is actually not a new paradigm in the context of mobile networks. Since for GLTE we have been developing more softwareized virtualized solutions. Sdn NFVs emerged more and more and more compute was integrated into the networks in themselves to be more flexible, to be faster and also enable maybe faster updates. With 5g. However, the edge itself, or edge computing itself, started to play a different role because we have started to look into a different set of use cases and requirements coming from them. The so called verticals, such as autonomous driving, health care, robotics, gaming related applications that do require extremely high reliabilities, but on the other hand, also very low, latencies. And in this context the mobile edge cloud has been started as a concept by Etsy which was then renamed as the multiassist edge which basically brings in the application closer to the end users for the purposes of reliability and latency. And this led to quite a bit of changes in the 5G networks itself, the mobile network architectures themselves. It did not only mean that the Edge was basically an attachment to the core networks or radio access networks it was really an integrated attempt to pursue and any changes that were required in this context for the core and access networks have been introduced with five G and it is continuing to evolve further with 5g advance and in the context of six G. Now, if we talk about 2030 and beyond, we are really looking into the actual convergence of computer networking and what this means for both the network side as well as the compute side.

[07:04] Jo Peterson: Well, thank you for all that color, that is amazing and it is very clear to me that edge is a term that means different things to different people. It's complex and it also seems to me to be a term that is continuing to evolve and mature. And speaking about maturity Meryem, if we are looking from a technical standpoint, how mature are we right now with edge connectivity on a scale of one to 10 with ten being hey, we've all adopted five G and we're ready for six G. Where are we at right now?

[07:44] Meryem Simsek: Very difficult question, Jo, because 5G has just been started to be rolled out. So clearly concept wise, we have developed novel concepts and evolutions and technologies for 5G but 5G is not completed yet. On the other hand, we are asking ourselves what's six G going to be? What is coming in 2030 and beyond? And honestly it will very much also depend on the applications that we are looking at to validate and evaluate the maturity for those. Can we really guarantee the service requirements? Can we enable seamless usage of compute and networking? This, this really depends on the applications that will emerge. We do talk in the context of six year also about the merge of digital and physical worlds and applications that might emerge in this context. And consumers here can be really us, the individuals, but also enterprises, as well as industry and machines or robots. And it very much really depends on the maturity itself, really depends on which use cases we are looking at, but also which use cases and applications will emerge for six G. However, I think we do see the trends that are evolving and we are preparing for them. And we have been doing this since 2014 ish for five G and we have started to do so for six G already today.

[09:36] Maribel Lopez: It's so interesting when you think as analysts, we've been talking about this for a while now and it's always a long lead time before when you start speaking about something and when there's actual adoption of it. We have to do all the concepts around it, we have to do all the technical specs, we have to do all the testing, we start deployment and then we have to figure out some new applications. And eventually, somewhere down the road, usually within ten years, we get to whatever we first started speaking about and you started to allude to some of the future as well. And we've been talking a lot about the network and the convergence of different technologies. We've been talking about IP, Ethernet, optical layers for a while and I'm wondering what do you see as future trends in the network that we should be thinking about? Is the network going to be built from disaggregated components? Are we thinking about a different way that we'll be constructing the network in 24 and beyond as we look at six G? Let's say.

[10:45] Meryem Simsek: Very good question and actually these are exactly the questions we are asking ourselves also within the team. And one notion that is emerging in the context of next G or six G is the concept of network of networks. Meaning we are not only looking into a single network that is enabling us connectivity or delivering the services, but we do think that multiple networks will be available and the efficient seamless let me call it stitching of individual network components for really seamless immersive experiences for every individual is going to be the target for this purpose. Yes, you mentioned routing. It is really an essential role optimized routing and the convergence of IP and optical network components but routing not only in the transfer domain but actually also user sessions if you have different networks from different providers, how do we enable the continuity? And these of you are also some of the questions that we are targeting to look at and that is in the context of six g also emerging with the tight consideration of the convergence you mentioned. Technology is converging but also businesses are converging. I think it is a very exciting era nowadays to see also in the context of edge. Yes, telcos are talking about edge but also web scalers are talking about the edge and that convergence is going to be also very exciting time to see and from the technology point of view I think we can look into fairly heterogeneous, adaptive, distributed multilayer approaches and solutions regarding the business. That is something to be figured out probably in this new era.

[12:55] Maribel Lopez: Some of what you said just led me to another question and if we think about where we are today with 5G, does this approach to the new network start with things like the network of networks you were discussing? Start with network slicing. I think
you have it that way.

[13:12] Meryem Simsek: That's one way, yes. Network slicing in the broader sense not only for the ran and core parts of the network but really slicing also of for example compute resources or really end to entry sources is clearly one direction to look at, but not only for a single network. How can we enable and achieve slicing through multiple networks by using different components of different networks in a network of network environment. It's probably also a way to look at for the network sicing.

[13:54] Jo Peterson: Dr. Meryem if I think about it from a different standpoint and I think about the end goal being interoperability and performance telemetry across networks, what role does Open APIs play in that equation?

[14:13] Meryem Simsek: Clearly an important role. Generally Openness will also play an important role. Nokia itself has a network exposure solution that really provides a robust platform for creating new services by consolidating APIs and presenting unified access to the API framework for developers and third parties. And this is to enable secure exposure of network services towards third parties. We talked about a heterogeneous ecosystem that is emerging where security and assurance will play or privacy guarantees will play an important role, but also it's enabling a developer environment in SDK for operators and the community. And the service Mashup for creating really end to end offerings by combining any of the network assets into your application is really the goal of this API framework, which will really play a big role. Talking about Openness, I think we have realized this also in the context of not in the context of 5G, but in parallel to 5G. Open, Rand. Discussions have started. Open Grady access networks, open solutions, opening up interfaces for more innovation and flexibility have been initiated and clearly open APIs and network interfaces
will play an important role in the future as well.

[15:57] Maribel Lopez: Thank you. I think that we've all been looking at how opinion APIs might help us move the ball forward. As I say. So one of the questions I had, and tell me if you think this is relevant, is how realistic is the end goal for this next generation Metro and Ed network? What's the role of that in reducing operational costs, minimizing hardware dependencies and ensuring faster updates and upgrades? I know we've spent a lot of time talking about creating fairly static fixed architecture and we moved from the fixed proprietary concept to an open public cloud concept and then we kind of pulled it back a little bit to be hybrid private when we thought about cloud computing. And I wonder how you're seeing this happen in the Edge. What are your thoughts around that?

[16:51] Meryem Simsek: I would like to see it to fault this question. One is from a technical realization point of view. The other one is probably with respect to how realistic this end goal is from the business point of view. Technically, we are actually entering a fairly distributed environment, distributed mesh type environment for edge networks, that is with multiple providers actually so fairly heterogeneous and distributed. And we are talking also in the context of next g about distribution and decentralization. There is a but here, which is the more we distribute, the more at some point we have to bring these things together, which might be, yes, we are distributing, but then the centralization of this is causing more workloads and more overhead than actually looking into maybe more less distributed solutions. So technology wise, technically we do have to find here the right balance. But I am optimistic that if we find the right solutions that operational costs as well as hardware dependencies will be reduced significantly from the business point of view. I must say, I'm a scientist, so I'm not a business expert. It very much depends on agreements as well as future business models where I can't comment too much on this. Generally, I think we are seeing trends towards the Edge network, be it the Metro or the edge itself. There is a notion in the context of telcos named Far Edge also even closer to the access or end users, we do see the trends and we are developing effective and high performing solutions for those. So I'm very optimistic to reach this end goal from the business point of view. I leave this to the business experts.

[19:04] Maribel Lopez: Makes total sense sales at that time.
[19:09] Jo Peterson: Yeah, and I just learned a new term today, dr. Meryem Far Edge. That's a new one for my lexicon. So I'm going to look that up thank you for that. So we always end with a fun fact, and it doesn't have to be technology related. So we'd like you to share a fun fact with us. Dr. Meryem.
[
19:31] Meryem Simsek: Yeah. This is maybe the most challenging question, I must say. I have been thinking a bit, and I I just recently joined Nokia and and looked into their history also. And Nokia has been founded in 80, 60, 518, 65, so in and the 19th century. And it was a paper mill, and one of their first products was toilet paper, which was a big surprise to me. So innovation can happen everywhere. That's my conclusion of this.

[20:07] Maribel Lopez: Love it. We're actually I was always amazed at the history of Nakia and how it continued to reinvent itself. And I think it's a good lesson for industries at large, right, that it can happen.

[20:19] Meryem Simsek: Absolutely. Yes.

[20:21] Jo Peterson: And I love that you have a sense of humor, Dr. Mario. That is so cute. Thank you so much for being with us today. You were so enjoyable, and it was great to sort of get your thought leadership on what's coming next. That was pretty cool. So thanks for spending time with us.

[20:40] Meryem Simsek: Thank you so much. Jo and Maryville was a pleasure.

[20:45] Jo Peterson: Thank you.

[20:46] Maribel Lopez: Take care. Bye.