CSPs, AI and cloud migration: A conversation with Amdocs

This article first appeared in Silverlinings.

The telco sector is readily embracing cloud. While many companies are already into the advanced stages of their cloud journeys, some will be grappling with the complexity of transitioning existing applications and data.

Indeed, it can be a complex beast. In this video interview, we sit down with Stephen Ellis, Division President and General Manager of Amdocs Cloud, to talk about AI driven intelligent cloud migration.

Can AI be a game changing tool if it enables telcos to streamline cloud migrations and transition applications seamlessly?

According to Stephen, communications service providers (CSPs), in general, are still in what he describes as ‘the early innings’ of cloud adoption, with the majority focused on low hanging fruit such as rehosting infrastructure.

However, the story is shifting to application modernization, which is where the real benefits offered by cloud are beginning to be realized.

Here, Stephen says, AI can play its part in the planning stage by assisting in the discovery and dispositioning of the applications portfolio. Likewise, when it comes to executing the transition to cloud, AI can also help by automating several of the critical processes which are primarily a human-led effort.

The conversation continues with a further discussion on the use cases for AI within CSPs, from regulatory compliance through to whether it can level the playing field with the cloud hyperscalers.

Tune in to watch the full discussion!

Steve Saunders: Hey there, I'm Steve Saunders on Cloud 9 and today we're going to be talking about AI-driven intelligent cloud migration. That's a big and important topic in our industry right now. Almost all telcos have already begun their cloud journey, but transitioning existing applications and data can be really complex and difficult.

AI could potentially be a game changer in the future if it delivers on the promise to streamline those migrations. AI-driven cloud migration could have the power to revolutionize cloud adoption, leveraging automated tools and advanced capabilities for faster, more accurate results.

With AI's intelligence, more applications could be refactored and replatformed, which will help increase the value of cloud adoption by enabling true elasticity and agility for migrated applications. So, all of this could mean increased efficiency, value, and ROI, making the cloud business case significantly more compelling.

With me today to discuss these issues is somebody who's superbly qualified to talk about them - Stephen Ellis - who is the Division President and General Manager of Amdocs Cloud. Stephen. Thanks so much for joining us today.

Stephen Ellis: Hello, Steve. Thank you for having me.

Steve Saunders: I wonder if you could perhaps provide an overview of the current landscape regarding the adoption and transition to the cloud within enterprises.

Stephen Ellis: Sure. I think there's a wide diversity of where CSPs are in their cloud journey. You will have heard lots of marketing-making commitments to cloud with the hyperscalers. But in terms of the actual transition to cloud, we're still in the early innings of adoption.

CSPs have focused on the early wins, which is primarily the infrastructure rehost, lift and shift use cases. But now as those things have started to complete we're pivoting our attention into more of an application modernization story. These are applications that enjoy some benefit to leveraging cloud-native capabilities and frankly, they're a little bit more difficult to get into the cloud.

As we pivot into more of the modernization story, our CSP customers are finding more challenges around how do you select the right applications to modernize in the cloud? What's the business case for actually doing that piece of work? How much effort is that work going to entail to achieve the benefit?

There's a lot of up-side for the cloud story as you start to shift your perspective from rehosting and infrastructure upgrades to the application upgrades. But folks are just starting to get into this problem set now.

Steve Saunders: What about AI? AI comes along and how does that change the world for telcos as they embark on a cloud journey? Is it good news?

Stephen Ellis: There are two areas. Clearly, our customers and CSP’s specifically are doing data modernization to unlock AI capabilities within their firms. When it comes to the application modernization side of the house, there's a lot of technology that is available to help accelerate the modernization of those applications by leveraging AI, either in the discovery and the planning parts of your modernization or in the actual execution and the transition of those technologies into cloud native capabilities.

I think you're going to see an evolving set of tools that will be primarily driven through AI capabilities to assist in that modernization effort.

Steve Saunders: Just talk to us a little bit more about those tools. Can you put a little bit more flesh on the bones, as it were. I mean, what sort of things is Amdocs thinking about developing or developing for that matter, in R& D right now, that will help telcos with what is frankly quite an anxiety-inducing and overwhelming shift, isn't it?

Stephen Ellis: I think the goal is “how do we get more value out of our cloud investments?” When you look at the infrastructure modernization approach, there definitely are benefits, but as you get more cloud native with your applications, the expectation is to get more and more value from the elasticity and the cloud-native capabilities and services that you may not have on prem. Those sorts of things should be unlocked.

When it comes to how AI fits into the story, you have different tools for different parts of the lifecycle of that mass modernization journey. The first step is to analyze your existing portfolio of applications and figure out what is the right disposition of those applications into the cloud.

Some of them are going to be rehosted. Some are going to be re-platformed. Some will be refactored. Some of them will be replaced. AI can assist in the discovery and the dispositioning of the portfolio of applications. And AI can also help you figure out the best path for each of those applications into the cloud.

AI can help us actually evaluate what this application does, its’ modernization path into the cloud, its disposition through the recommended ”R”, and how to actually estimate the effort and the benefit that would come with moving that application to the cloud according to that “R” recommendation. That's how AI can help us from a planning and discovery perspective.

AI can also help us during the transition into the cloud. Once we figured out what we want to put into the cloud and how we want to put into the cloud, someone's actually got to go in and do the work. Today, that's primarily a human-led effort. However, we're seeing tools come to market that start help automate this more. One of the large use cases that I see when it comes to modernizing is how do you split monolithic applications up into microservices? How do you containerize those applications? Right now, that's very much a handcrafted exercise. But we are starting to see tools that can figure out how to break apart these monoliths and start to auto generate the microservices and the container code.

Another use case that I see a lot is the mainframe. As customers start to look at their legacy mainframe system, they don't want to be in COBOL anymore. They don’t want to be in these antiquated programming languages. How do you get these things into modern programming languages using AI and code language translation tools? Everyone's heard of ChatGPT and the ability of ChatGPT to do these types of things. I think you're going to see more and more types of tools come out that assist in the conversion and the transition into cloud native.

Once you get the applications finally into the cloud, there's the validation side of the house. How do you make sure what you've transitioned to the cloud actually works? How do you automate the regression testing? AI can play a huge role in helping us automate the testing frameworks as well.

Steve Saunders: Let me ask you a question. I mean, there's a split in the world right now. On the one side, we have traditional service providers. On the other side, we have hyperscalers. They come from different universes. Hyperscalers, you know, they're quite arrogant. They don't really feel the need to interoperate with others. They offer very significant benefits, of course, and they're growing, you know, rapaciously. On the other side, we have telcos and service providers, carriers, and they have a different set of strengths. They have a very high level of trust with their customers. They understand reliability. They understand customer service. They understand how to run global networks in a very reliable way.

Do you think AI is a technology that can perhaps help service providers compete at a global level with hyperscalers or are these two different, and are these always going to remain two different animals in the communications universe?

Stephen Ellis: We definitely see overlap between where hyperscalers want to go and where CSPs are looking to go. We see many of our customers trying to build their own at-scale clouds. And I guess the question is, can AI help these folks compete? I think there's lots of room in the market for a variety of solutions when it comes to cloud.

I see spaces where CSPs can play large roles like in clouds that need to be geographically defined - when certain data has to live within the national perimeter, and a hyperscaler solution may not be the best fit. So a local solution may make more sense.

I firmly believe that AI can help all enterprises. We know that the hyperscalers are ahead of the game. They're helping unlock AI capabilities for all enterprises, but they've also been leveraging AI for their own enterprise for a long period of time. They've got the benefit of being ahead. But that doesn't mean that the CSPs can't catch up leveraging AI as they start to define how they want to be in business and whether that's competing with the hyperscalers or leveraging the hyperscaler services to build their own business.

Steve Saunders: And do you think that regulatory and compliance could be one area, particularly in the financial community, where AI could be useful? It seems like an obvious play to me - making sure that people are following their local and geo compliance rules. Or do you think that's still going to stay with the human operators because of trust issues?

I guess that kind of depends on where people are with their comfort level with AI.

Stephen Ellis: I think that for technology, compliance and regulations - especially when it comes to cloud - getting that framework right is going to define how you adopt cloud. You have to be compliant. You’ve got to be in line with the regulatory regime. I don't believe hand-crafting that is the right answer. Everyone builds compensating controls into their cloud platform and they unlock the services that make sense for them inside those frameworks. But I think AI is a huge part of that. AI is how you monitor how your cloud is being used and what are the rules for your enterprise when it comes to how you want to leverage that cloud. The observability of your cloud consumption and what you do with that data - that is a prime place for AI to play. And I think it's a pretty common design tactic to leverage AI for exactly those kinds of workloads.

Steve Saunders: You know, most new technologies - and it doesn't really matter where they sit in the network - suffer some sort of backlash after the initial excitement, don't they? You're in charge of a really important critical division of Amdocs. Now, when you look at AI, are you worried about being considered part of this? We had green-washing. We had cloud-washing. Now, are we going to have AI-washing? Do you worry about its capabilities being overhyped? How do you keep your feet on the ground in a cloud world and guide your customers appropriately?

Stephen Ellis: I think we have to focus on where the value is. If AI can help us, that’s fantastic. One of the terms you're going to hear a ton about in the market today is ChatGPT. You hear Gen AI. ChatGPT. It seems like every solution has some influx of Gen AI.

For me, it's about focusing on what the business objectives are and then picking the right set of tools for the job. A lot of my job today is helping our CSP customers tackle those very complex core BSS and OSS solutions, and figuring out how does moving to the cloud benefit the organization through agility, through auto-scaling, through reliability. Sometimes AI can be helpful for those transformations. But sometimes it's just core cloud capabilities. Like, how do I break these monoliths apart? How do I leverage modern cloud first architectures to deploy these technologies? And if AI can help get value out of these things - fantastic.

Now, I do see a lot of use cases where AI is super relevant: How we engage with the customers. How we leverage AI to measure why customers are contacting us, and how do we automate better responses back to achieve better customer satisfaction with our interactions. There are lots of natural places for AI to have a role.

I haven't really heard of any backlash yet. We're in a regulated environment. We're not going to be able to do things that are not compliant with our regulators. I see most of my customers trying to figure out how to leverage AI to extract value out of their cloud investments. I don't really see a slow down there.

Steve Saunders: It seems like AI is going to do more than just make things a bit easier and faster when people are migrating applications to the cloud. It's actually going to change the way they do that fundamentally. But how long is it going to take before we see that transformation take place with AI really taking an integral role in those migrations?

Stephen Ellis: It's not too far off. When you look at the set of tools available to us today on the discovery side of the house, we're already leveraging tools like Cast or CloudHedge to scan the GitHubs and figure out what kind of applications are in there, how they're architected and automate some of the disposition of these things.

We're not far off from a discovery and a planning perspective. I would imagine we'll see some evolution there. When it comes to the actual modernization efforts, you're seeing lots of great tools like co-pilot from Microsoft helping developers code or develop.

We're seeing lots of tooling in the language translation space when modernizing antiquated programming languages to modern ones. The state of the art's there. It's happening. Is it all packaged into one cohesive cloud migration framework? Perhaps not, but the bits and pieces of the tools are out there.

As we take on more complicated workloads to bring to the cloud, the more automation that we can bring to accelerate and bring consistency to our cloud migrations, the more you're going to see big CSPs adopt and modernize applications. And hopefully what you'll end up seeing is “modernized” becomes the predominant disposition for those applications as opposed to just rehosting.

Steve Saunders: Where do you see - this is a good interview question - where do you see Amdocs in five years’ time? You're at the tip of the spear right now with your job as general manager of Amdocs Cloud. No pressure.

Stephen Ellis: No pressure. We're helping our customers take their core BSS and OSS systems to the cloud. We want to pave a path where our clients can get into cloud-native modern architectures that scale and give them all the agility and flexibility that they need to scale their business.

That journey is a multi-year journey. And that journey may not mean jumping into the brand spanking new technology right away. It may be a gradual modernization and a gradual transition. Our job over the next five years is to get our customers on the path; get them building those roadmaps, evolve the tech stacks that they have today and build the bridges that unlock the gradual modernization into those cloud native technologies so that we can leverage the cloud native capabilities that allow us to autoscale, that allow us to run resilient architectures with no downtime, but then also to unlock the capabilities that data and AI bring to the table as well. And how we leverage BSS and OSS systems to interact with our customers and drive more value.

Steve Saunders: Yeah, that's a very exciting time. You have an interesting job and I like the pragmatism about what you're doing at Amdocs with cloud and AI. And I'm sure your customers are very happy that you're taking that approach.

So Stephen, I really appreciate you talking to me today. It's an incredibly exciting, fast moving and dynamic area. And I hope we can stay in touch as Amdocs continues its good work in this space.

Stephen Ellis: Thanks for having me here today. Telcos are moving to the cloud at scale and at pace. And Amdocs is here to support them.

Steve Saunders: Thank you, Stephen.