Sean McManus, TelecomTV (00:05):
Hi, my name is Sean McManus from Telecom tv. I'm at the Open Networking and Edge Summit, and I'm joined now by Inderpreet Kaur and Chris Sarantopoulos, both from Omdia, and they're going to tell us about some of the latest research. So int tell us, first of all, telcos are increasingly on a journey to virtualize and containerize their network infrastructure. What's driving that?

Inderpreet Kaur, Omdia (00:27):
So before we get into the drivers, I would really like to highlight what this cloud native is because we've heard about this term a couple of times and there are three benefits we are looking at when we talk about cloud native. It has to be flexibility, scalability, and resilience. So when I say flexibility in an old world of telcos, even with the virtual environment, you build the system for a set capacity, but that's not the case with cloud native. I'm a telco, I see a certain search in traffic and I want to have capacity on demand. This is what cloud native allows me to do. I'm a telco, I'm serving a dynamic slicing use case and I want to have a slice or I want to have instance of the network function at the edge or the far edge of the network. Cloud native gives me that option.

(01:26):
I want to do it on-prem or I want to do it on a public cloud infrastructure. Again, that's the choice that cloud native brings me. Scalability, it is really about how dynamically you are managing those cloud resources. Again, if we look at Kubernetes, it has done that for the enterprise environments. You can scale in, scale out the capacity on the go, on the fly. It is a declarative mechanism and you add the GitHubs related frameworks to it. It adds the additional automation capabilities to your resource management. Then you talk about resilience. Again, it's about how you recover your entire network state or the network service in case of a failover or a failure. Again, Kubernetes has done that. In enterprise environments. You can auto heal or auto scale your clusters in case of a fail cluster failover. So these are the benefits I think cloud native has to bring in for telcos when they talk about CNFs.

(02:37):
We do this research every year and it's a end user perception survey wherein we ask, go and ask the telcos, where are you on this adoption journey? So every year we get the results and we tend to compare the results we get this year with the previous year. So the latest results that we have, telcos indicated that about on average 20% of their workloads are in cloud native environments. Now look at the previous year's surveys, it was about 16%. So it's not a massive jump, but there is some progress. But at the same time, if I look at where are the networks in terms of virtualized or virtual machine kind of environments, a good 35 to 40% of the network functions still sit in that environment and what the survey indicates is that that is likely going to be forward as you go forward five years as well. So maybe about 30% odd network functions will still sit in the virtualized environments. That's I guess where we are reading the industry or the journey so far.

Sean McManus, TelecomTV (03:47):
What do you think the reasons are for Telco sticking with virtual machines to that extent?

Inderpreet Kaur, Omdia (03:52):
I think there are multiple reasons. I can call out three of them. First is telcos have been operating in this environment for a very long time, so definitely they're more familiar with the skills that are required, the frameworks that are being used. The second is in certain environments, VMs can bring in better security compared to containers. And the third thing is more related to you've invested in the networks, you've been investing in virtualized environments for past five to eight years. You really need to sweat that infrastructure before you can go out and scale the cloud native environment. So I think that's the journey where it is progressing with virtual machines and cloud native environments. But again, to call out, when I look at how do or what proportion of cloud native network functions telcos will have by 20 20, 20 30, I think it's 33% or 35%. That is about the result that we saw in the recent surveys, and I compare it with our previous surveys. I did not see a lot of variation in that percentage. So it kind of indicates that not everything can be containerized and neither is the case that everything can move away from the physical world into a virtualized or containerized environment. So there will always be this balance and telcos will maintain this mixed kind of environment where you have certain functions in physical, certain in virtual and certain in the cloud native environment.

Sean McManus, TelecomTV (05:32):
It was interesting to see that telcos are increasingly using the same cloud infrastructure for their IT and their network functions. Tell us what's motivating that?

Inderpreet Kaur, Omdia (05:42):
Well, when we read about different telcos adopting or moving towards a horizontal cloud infrastructure, there are different telcos taking different approaches. I would say this horizontal cloud infrastructure is what we relate mostly with tier one operators who have multi-country footprint and would want to have a similar kind of cloud, a consistent cloud environment across different countries across their footprints. So that's one, but to really have a common cloud infrastructure, it's about how well you are managing your resources. So traditionally telcos have maintained separate environments, a different cloud for using or hosting your IT functions, oss, bss, and a different environment for hosting your network related workloads. Now there's a lot of phase stage of resources when you are maintaining separate environments. So if you have a horizontal environment, it definitely helps you utilize those resources better.

Sean McManus, TelecomTV (06:45):
Thank you. Chris, tell us how advanced automation is in CSP networks today.

Christoforos Sarantopoulos, Omdia (06:51):
Well, to define this, first we have to come up with a common metric across the whole industry that defines the status of automation and the level of automation in an industry as a whole level. I'd say that the team forums, autonomous network levels model is establishing itself as the defacto measurement for the status of automation in reality and an internal level, we see CSPs working their own ways. So we see that most of them, the 45% according to our research, are using their own metrics to define the status and the level of their automation, their networks, and around one third resorting to the TM forum autonomous network levels, the rest will don't have a metric at all. That was surprising at this point. To answer your question, I think based on the team forums, autonomous network level's model, we see that more than half of the CSPs are level two or level three.

(08:01):
So this is where AI is starting to getting more consulting role. So it's making suggestions to operators, but still the decision is made by humans and human operators. We see also that the I-P-N-N-P-L-S are the most automated network domains, and this kind of makes sense. These are where the technology is not telco exclusive. So it is used in enterprise networks and more importantly, it is used in data centers that lead the innovation in the automation at least. And I would say that though, however, regardless of the domain operators should look to find out what are the most high value use cases that they need to automate, find these high value scenarios that are a combination of the business value and the operational feasibility. So how easy it is to use the technology to do things like service assurance and fault management are low hanging apples in this context, they're more advanced technology available to automate these domains and also they provide very quick wins because they immediately impact the customer experience. On the other hand, things like network planning or network rollout, our use cases that are heavily manual, they don't care that often. It's not an everyday activity and it also, it's not a huge burden to A CSP topics. So it makes sense that they are not that high priorities.

Sean McManus, TelecomTV (09:41):
Tell us, what are the biggest challenges that CSPs face with regard to network automation?

Christoforos Sarantopoulos, Omdia (09:47):
Our research shows that by far the two biggest challenges is integration and interoperability of network systems and then employee skillset gap. The first one is that because telcos have a huge legacy burden, they rely a lot on legacy systems to run their networks. The systems are not designed for modern automation. They are not API based and it's difficult to use them or to integrate them to automation solutions. Also, there are many industry standards that try to define the framework for automation within telco networks, but in many cases they are very open to interpretation. So many vendors choose or create their own implementations and this leaves a gap for CSPs where they either have to come up with and develop some customized solutions to accommodate that or resort to some middleware to ensure interoperability across the board. The second one is the employee skillset gap. Telcos excel in traditional network engineering skills, but they have a gap on a shortage of available talent in terms of skills that are more software driven, are better used for modern technologies. So for example, to implement infrastructure as code engineers need to be able to script a bit and use tools like Ansible, Terraform, and this type of tools to successfully manage the 5G core operators need to work with Kubernetes. So being able to manage network resources, you're using helm charts. This is with things that where they need to find the right mixture of skills and create the teams that are also not siloed, but bring together network engineers with traditional skills and knowledge of the with more software driven skills that can manage these new technologies.

Sean McManus, TelecomTV (12:01):
I was interested to see that about a quarter of your respondents said that budget and a lack of certainty about the return on investments were a key challenge for them. Where do you expect the return on investment to come from?

Christoforos Sarantopoulos, Omdia (12:13):
So telcos basically have two levers to pull, right? So they can either reduce the costs or they can either maximize the revenue. This is obviously the benefits. This automation always touches everything across the board. So at the beginning, the operational benefits and the reduction of operational costs is the scenario that can create, have a more direct impact revenues, new revenues will need other elements like dynamic services. This is something that is coming in the recent years. Things like new enterprise services or cloud services, these are the types of scenarios that will create, generate a new revenue for telcos to justify this automation spending. Now, in reality, when it comes to decisions over budget, we see that creating new revenue takes a precedence and the decision makers will always have in the bottom priority the facts that create operational efficiencies. But in reality, these are the two cases and we see it a lot whenever an operator sets up a task force for autonomous networks, the first thing they want, the first task they say is find me or quantify the benefits for me, and this is the most difficult part to do. Right?

Sean McManus, TelecomTV (13:51):
Absolutely. What would you say was the most surprising thing you found in the research, Chris?

Christoforos Sarantopoulos, Omdia (13:57):
So I would say that there is still a huge gap in terms of the tooling that is used to automate networks and the approaches that the operators take so many operators resort to. There are many approaches, obviously many operators tend to rely at the moment to their network equipment, equipment vendors, so they don't feel comfortable using a custom made solution or developing their own solutions there or going to a vendor that comes from a different background. I think this is a bit of, not a surprise as per se, but I think it shows a trend in telcos.

Sean McManus, TelecomTV (14:47):
Thank you. Inderpreet, what was the most surprising thing to you?

Inderpreet Kaur, Omdia (14:50):
I think we called out that horizontal cloud adoption, and when we did the surveys this year, we asked this question to the CSPs, have you adopted the horizontal cloud for it? And the network functions huge proportion. 19% said they don't have, so 80% said they already have, or they are working on implementing a horizontal cloud strategy. So I think I would like to call that out. Now, this horizontalization could have been add different layers. So it could either happen at the infrastructure layer or the Kubernetes CAS layer or at the upper level where you have the management tools. So most of the telcos are still at the lower layers where you have either the CAS or the infrastructure, which is a common platform for your IT and network functions. But a good percentage, I think 32% of those respondents said they also have a common platform for managing those resources. So it could be a cluster for your network function or a cluster for your IT applications. They can have common tools to do that. So that was surprising in our research

Sean McManus, TelecomTV (16:11):
Inderpreet and Chris, thank you very much.

Inderpreet Kaur, Omdia (16:12):
Thank you.

Christoforos Sarantopoulos, Omdia (16:13):
Thank you.