On this episode of The Six Five, hosts Patrick Moorhead and Daniel Newman are joined by AWS‘s Holly Mesrobian, VP Serverless Compute for a conversation on the integration and future of AWS Serverless Compute, including Lambda, ECS, Fargate, and Event-Driven Architectures.
Our discussion covers:
- Navigating Kubernetes complexities with AWS Serverless
- Celebrating 10 years of AWS Lambda and its impact on customers
- How AWS Serverless changes application architecture towards event-driven models
- Integration of applications within the AWS portfolio akin to iPaaS
- Meeting the demands of Generative AI workloads with serverless solutions
Learn more at AWS.
Watch the video below, and be sure to subscribe to our YouTube channel, so you never miss an episode.
Or listen to the audio here:
Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Transcript:
Patrick Moorhead: The Six Five is back. And we’re talking serverless technologies. Dan, it is great to see you, my friend. This conversation is going to be great because we have talked about AWS Serverless for the prior two years at re:Invent 22. We talked about serverless at the product management level, and then at 2023 re:Invent we got talk to Nick Kuhl, the GM. That’s amazing. We are working up the stack here. We must be doing something right, I guess.
Daniel Newman: Yeah, it’s a good time, Pat, for a stack joke, because when you talk about development, you got to make a joke about working up or down the stack. But yeah, look, anytime we have the chance to talk about the development of applications and then all the important integrations that take place between developers and of course infrastructure, it opens up the door for a wonderful conversation. We’ve had many with the folks at AWS and I’m excited about another one here today.
Patrick Moorhead: I know, and that thing seems like a great time to bring in Holly, the Vice President of Serverless Compute. Holly, great to see you.
Holly Mesrobian: Thank you. It’s great to be here. And I’m going to add my own stack joke. We’re going to pop the stack.
Patrick Moorhead: Ooh, that’s going to be good. And I’m going to have to relay that to my son and be like, Hey, what do you think? Share that with all your buddies in college. No, but seriously, it’s been a really interesting time here and discussions with you. And typically a lot of those discussion have been separate. We talk about ECS, talk about Fargate, Lambda, event driven architectures, really looking at it as distinct and separate services. But now we’re here to talk about the strategic alignment, strategic significance of AWS Serverless as an operational model.
Holly Mesrobian: That’s right. And you’re here talking to me today. And those services have all been combined in under my leadership. And so I work across the space with those leaders every day. And it’s something that we really talk a lot about, is how do these products work side by side. ECS and ECS Fargate in particular is a great container orchestrator for serverless operating model. Lambda has been well known for its function as a service. And then, of course, they operate really well with EventBridge and Step Functions. And so we’ve combined these all together and my team sits down and thinks about this deeply on an ongoing basis.
Patrick Moorhead: By the way, I think there’s a collective cheering in the customer section as they’re watching this. And I’ve always wanted to see the collective nature of it. And some customers love singular products, but as I think we get bigger, as enterprises get bigger and bigger and they increase their capabilities, they want to see a holistic strategy across all of these. So thumbs up. And we’re only two or three minutes in here so far.
Holly Mesrobian: Oh, that’s great.
Daniel Newman: Yeah, my instinct is ask her a question about that mountain of books behind her, but I don’t know if we have enough time. So what about a question about Kubernetes? How do you feel about that Holly?
Holly Mesrobian: Yeah, so we support Kubernetes and EKS as well. We try to support customers in what they’re trying to achieve. And some customers do choose Kubernetes as their model, but we also have our serverless model as well. And with ECS and Lambda and Step Functions and EventBridge we feel like we have a lot of flexibility to really help customers build more faster.
And when I think about what are our products, we have a vision inside of our product portfolio and it’s about delivering availability and security and cost efficiency for customers better than they can for themselves. And really the reason why we want to do that is to remove all of that off a customers so that they can build quickly and bring those ideas to production and then really rely on us AWS to deliver all those fundamentals underneath for our customers. And so they’re just not having to think about that on a day in day out basis.
Patrick Moorhead: I love it. Dan just jumps right in there-
Daniel Newman: Well, I was going to dig in on that, but she did such a good job. Sometimes it’s like I think she read my mind. I was going to be like, tell me about AWS Serverless and Kubernetes. All I had to say was Kubernetes and it was like boom. And then off she went.
Patrick Moorhead: She’s literally off to the races. This is great. So I believe this is the 10th anniversary of AMD Lambda. Happy birthday. I hope you got Lambda a birthday cake and everybody blew out the candles or Lambda blew up. I don’t think there’s somebody named Lambda, but anyways. Anyway, just work with me on this one. So serious question coming up. How has Lambda positively impacted your customer and workloads? What is the metric that they’ve looked at? Is it a faster time to market? Is it simplicity? Is it something different?
Holly Mesrobian: Yeah, so it is a faster time to market and then also getting those underlying fundamentals that I was talking about without having to do the engineering themselves, which of course helps drive that faster time to market. In terms of metrics, the ongoing growth with millions of active customers and 10 trillion or more invokes per month is really showing the growth of our customers on Lambda.
And that’s the 10 year anniversary. You’re going to hear a lot about it throughout the year because we’re really excited about that as well. But we continue to just drive innovation back into Lambda and we’ve expanded the product and the product capabilities on Lambda every single year. And I’ve been with Lambda for more than six years now and we’re just working on that all the time.
Patrick Moorhead: It’s interesting that the new stuff is always great and 10 years ago, obviously Lambda was new, but also something having longevity is good. And I think a lot of enterprises before they jump into something, like to see something that’s been tested by many customers, having billions of invokes. And I’m sure that’s not annually that there’s a lot bigger number for that, but it gives them a peace of mind.
Holly Mesrobian: Yeah. And we have seen that in terms of how customers tend to start working with Lambda, a lot of times they will start off kicking the tires, they’ll use back office jobs. It’s a classic use case. But then we see customers over time who move to really that serverless operational model or serverless first. And what we’ve seen is this pivot from the old days where it was merely like back office jobs to now it is their primary way of developing their primary operating model.
And they come in and they’re like, okay, I’m going to build everything serverlessly because of the operational model that they like and they’ll build it all on Lambda. And then if it’s not a fit for Lambda, then they’ll go back and they’ll build it on ECS Fargate. But it’s really supporting that lower operational model that helps them build faster and put less into day in and day out operations.
Patrick Moorhead: I know you love all your children the same, but I do believe ECS is celebrating its 10th anniversary too. So if you’re going to have a cake for Lambda, you must have one for ECS.
Holly Mesrobian: Well, maybe it’ll be a cake and a pie or something because you have to make each child feel special.
Patrick Moorhead: Definitely. You are definitely a parent. Yes.
Holly Mesrobian: Yes
Patrick Moorhead: Absolutely.
Holly Mesrobian: Carrot cake and chocolate cake, we’ll figure it out, but…
Daniel Newman: You’re not a middle child. I think that’s how it goes. So AWS Serverless, it’s changing. It’s the way applications are architected Holly. It used to be all about being confined to the infrastructure that you own, and now you can go serverless and you can be event driven. How are your customers starting to think about using this event driven architecture to develop applications differently? And what EDA services are you guys offering to meet their needs?
Holly Mesrobian: Yeah. So you noted my book collection back there and we can pull it back to software engineering principles, highly cohesive systems, loosely coupled event-driven architectures, all of distributed systems. And what you see is that customers are really relying on us to build a lot of those capabilities and they’re able to leverage them and reuse them, again, great software engineering principle, and then build the loosely coupled integrations in the highly cohesive code that stitches those systems altogether. And so what we’re trying to do is provide all of the systems like with Step Functions and EventBridge and then all the engineering best practices underneath Lambda or ECS. So that really it comes down to customers building their highly cohesive code that’s really unique to their business and not having to build all that boilerplate. And that’s really what we’re providing. We’re building and managing all the boilerplate and the best practices for distributed cloud engineering.
Daniel Newman: But this is something that’s really changing the game pretty significantly, in your role as part of the leadership, you’re speaking to a lot of the customers now, so you’re starting to get a sense, but everything from time to market to the ability to develop for this AI transformation. I’m sure we’re going to get to that later. Gosh, Pat, do you believe it? We made it 10 minutes into a conversation, I don’t think we’d said AI yet.
Patrick Moorhead: No, I know.
Daniel Newman: But I have to imagine that this event-driven architecture approach is going to become more and more important as companies are moving to more diverse infrastructure, diverse compute, and then different development environments. You guys are really enabling them by creating this technology that’s going to streamline it all for them.
Holly Mesrobian: Yes, and that’s exactly right. Again, going back to those event driven building blocks, and one of the things that we really hear a lot from customers is that you can easily pull out one of those building blocks that’s unique to their business and then plug a new one in and you’re not opening up and cracking your entire system’s architecture. It’s like you have something you’re trying to do, it’s very discreet.
And then say you’re building something that is a foreign application and it could be a GenAI related application, today you have it doing one thing, you pull that one thing out and you put that new thing in and it’s very easy to make this pluggable systems. And you can do that everywhere. And we also hear from customers who are in very highly regulated industries, and especially when they’re in highly regulated industries, the thing that you’re just pulling out one little thing and not cracking open the entire box helps them tremendously with their speed of innovation because they’re not having to go back and re-verify every single piece of a working system. It’s just one little piece.
Patrick Moorhead: So you have customers using the entire Serverless suite. Some of them might use an ECS or a Fargate or a Lambda event driven. How are customers or how does the portfolio integrate different applications together? Is it like people use with iPads and APIs? I know it’s not APIs necessarily like you would in iPads, but is it something similar?
Holly Mesrobian: We do see where, and if you think about the interface, a lot of the interfaces are APIs, and that’s really what you’re pulling away is put a simple API in front and then you can hook together your APIs.
Patrick Moorhead: Excellent-
Holly Mesrobian: Yeah-
Patrick Moorhead: … by the way.
Daniel Newman: Sometimes it’s quick and to the point. Look, we live in a world where more people watch TikToks. We’ve got to figure out how to get things out into the wild quickly.
Holly Mesrobian: It has to be 20 seconds.
Daniel Newman: We can actually take that cutout and just answer that question, how are they using it? iPads silly. It’s APIs. But I do want to, now I started you up in my last question, Holly, and I’m going to take us back there, but all right, GenAI, we know, and of course AWS is doing a lot of great innovation. Both Patrick and I have published a number of different research pieces and have both been very positive in terms of the overall AWS approach. But let’s talk specifically about serverless. If you’re trying to develop a generative AI workload and you require serverless infrastructure, how is your team, how is AWS helping your customers get that done?
Holly Mesrobian: Yeah, well, so I think there’s a couple of different pieces here. One is I will go back to it’s early and there’s a lot of iteration happening. There’s going to be a lot of changes in this space over time. And the ability to take apart and put back together and build applications really, really quickly is key. And then we continue to build integrations into our services. If you think about back at re:Invent, we launched a integration with Step Functions and Bedrock. And you’ll continue to see us look at integrating new services through APIs as the space evolves. And then last but not least is we’ll continue to look at, we do have a compute offering in our serverless and we’ll continue to look at how best we serve GenAI capabilities through compute as well.
Patrick Moorhead: So I have to ask, this might be one of these 15 second responses. So consider this a follow-up. Does it meet the bar for scalable and optimized serverless services when it comes to what you’re creating with GenAI?
Holly Mesrobian: You’ll have to explain to me a little bit more what you mean by scalable and optimal services.
Patrick Moorhead: Sorry, I guess what I mean is when a lot of enterprises are looking to create, or create a generative AI application, they want to make sure it’s fully scalable and they want to make sure it’s optimized for generative AI. And I’m just wondering is will those be checkboxes when you develop GenAI serverless services?
Holly Mesrobian: It’s certainly from the standpoint of scalable, that is one of the fundamental underpinnings of serverless and that we scale up on-demand as your workload demands is definitely like that. That is a starting point and it does tie back to those fundamentals that we were talking about earlier.
Patrick Moorhead: Excellent.
Daniel Newman: So what about, I’m going to pivot just a moment because with all the growth around AI, there’s another sub theme that’s become more and more in the mainstream and that’s security, Holly. When you’re thinking about more data, more automation, more artificial intelligence, we’re also creating more complexity to securing the environment. How are you and how was AWS thinking about cloud security as it relates and are you making it foundational to the serverless business?
Holly Mesrobian: Yes. So when I think about what do we do in serverless, and a lot of our customers, a lot of our customers come to serverless solutions because they see code as liability and they look at what we do in serverless, where we’re removing a lot of the boilerplate code. We’re taking responsibility for a lot of the patching and the management of environments. We provide guardrails and security best practices built into the services.
And so they really look to us to help reduce their exposure to security vectors and then build on top of us as a fundamental for them so that they have all those best practices and they say, okay, you’re thinking about it all the time. You’re looking at it all the time and you have engineers working on it all the time, which we do. That is what we do. And so we really do provide a really solid foundation for security, best practices, for customers building on our serverless products. And I think going back to the APIs, APIs certainly also help as well.
Patrick Moorhead: So I want to ask about platform engineering. Sometimes people look at platform engineering as serverless, is that the two shall not meet? And I’m curious, how do you see the co-mingling of platform engineering and serverless?
Holly Mesrobian: Yeah, so we actually look at our customers across a breadth. Developers are our history and developers have been customers of serverless, but we also see platform engineering teams inside of organizations really liking or appreciating the value that serverless brings too, because it helps them, they can configure the serverless inside of an organization of an enterprise, and then they are offering that to their developers with less work for them.
Patrick Moorhead: I get it. Thank you.
Daniel Newman: So what about the economics of it all, Holly? Look, companies right now are in this really interesting inflection as it pertains to the economics and the business. We talk about cloud as an operating model and then, of course, where everybody’s assessing what do I own, what do I consume? You got everyone going back and forth between subscriptions and licensing. Obviously I think we know where it’s all going, but we’ve seen there’s still a lot of economics being thought about. Serverless has at least what we’ve seen from analysts perspective, some economic advantages. And what are you and what are your customers telling you about the TCO advantages that are being created by investing and going serverless?
Holly Mesrobian: Yeah, so over time we have done a number of studies that show that serverless does have very good economics for TCO. And we really think of TCO as two things, both the infrastructure costs, which of course can play really nicely, especially when you look at, we take all the scaling up and scaling down, and if you look at engineering systems, a lot of times you look at peaks and troughs. And so your economic model goes back to, yes, you pay during the peaks, but you’re not paying during the troughs.
And so that works really well. And then the other piece is the long term operating costs. So your engineers, thinking about your engineers as an actual valuable resource, they can either go and build your business and build new applications, which is what we want them doing, or they can do a lot of the day in day out operations, which is a lot of the value that we’re trying to bring underneath the serverless operating model.
Patrick Moorhead: Yeah, it’s compelling. And I want to read what you have done on TCO on serverless. It does beg the question inside of AWS, you have VMs, you have containers, you have serverless. What is the adoption rate? What percentage of whatever metric you want to come up with, your clients or workloads are serverless compared to VMs and containers?
Holly Mesrobian: I think the interesting one isn’t outside of Amazon. I think it’s the inside. And we have a lot of our AWS teams that build on serverless and on our serverless container offerings as well as across Amazon. So we have very broad adoption inside. And if you look at our services, most of my serverless’s services in my portfolio are launched with every new region. Pretty much all of my services are, and they’re launched in every new region because customers demand them. But also because so many of our Amazon teams, our AWS teams build on them that they have to be there because we can’t launch a region any other way.
Daniel Newman: Well, Holly, we’ve definitely been enjoying this. We got a few minutes, I got a couple more things in me up my sleeve. So we’re doing rapid fire, we’re hitting you up with all the good questions. I want to talk about the ground truth of your business, and that’s what your customers are saying and how they’re benefiting.
We’ve talked a little bit about adoption and we’ve talked a little bit about TCO, talk about how it’s all working, but what’s the anecdotal benefits, the biggest things when a customer says the biggest benefits that I’ve had, okay, I’m a big retailer and I’ve moved more of my infrastructure, I’ve moved more to serverless, I’ve moved… When you talk to insurance, a financial services company, what are they telling you are the big, immediate and therefore longer benefits that they’re able to realize?
Holly Mesrobian: Yeah, I think you’d say financial services. We have a lot of financial services customers that use our serverless products, and a lot of them are in the mindset of serverless first. And the reason why they are is because it is time to market. And they actually, internal organizations, large organizations, a lot of times they do have ways of measuring their agility and how much their engineering teams are able to deliver over time. And it really ties back to that, it is the pace of innovation and the speed of delivery that they recognize as a real value add to their business, especially when they’re trying to move fast and stay relevant in a changing environment.
Patrick Moorhead: We’ve had a great conversation on the past, we talked about ten-year anniversary, talked about what you’re doing today. I’d like to talk about where you’re going and know, we’re not asking you to spill your roadmap, but you’re certainly welcome to if you’d like.
Daniel Newman: Cover your ears, comms people.
Patrick Moorhead: Yeah. So maybe give some insights on your North Star, maybe how you plan on differentiating, let’s say for generative AI, even though I know we’re in the early stages of serverless and GenAI. Yeah, I’ll leave it at that.
Holly Mesrobian: Yeah, so we will continue to build improvements in our products and support in generative AI. We see a lot of activity now in terms of model training and kicking the tires. What I personally am very excited about in terms of the serverless products is when those tire kicking turns into real applications, because that’s when serverless, the ability to build fast and do a lot of invention and proof of concepts and get things out there and working is going to really pay off.
Daniel Newman: Well, Holly, it’s been a lot of fun and we appreciate you letting us ask, hit you with question after question. You’re leading a really large and important team there at AWS and we look forward to following the progress. And guess what? You are the tip of the spear of a number of conversations that we are going to be having this year with a number of your teams. So we’re going to be talking about integration on AWS, disrupting legacy iPads.
We’re going to do that in April. We’re going to be talking about serverless security in June. We’re going to be doing a 10-year special on both ECS and Lambda, one middle of the year, one end of the year. And then we’re going to have a special episode here to talk just about AIML workload. So what we basically did was we let you warm everything up and then we’re going to go deep on the next five topics and we’re going to do that throughout the year. So I just want to say thank you so much, Holly, for being a great guest and for letting Patrick and I hit you up with all of these questions and educating the community.
Holly Mesrobian: Thank you. It was my pleasure. And I can’t wait for you to speak with the rest of the team and hopefully all those conversations are enjoyable, like this one’s been.
Daniel Newman: All right.
Patrick Moorhead: Thank you.
Daniel Newman: Well, Holly, thanks so much and everyone out there, thank you so much for tuning in with us. Please hit that subscribe button, join Patrick and I for all of the episodes of the Six Five here with the AWS Serverless team. And of course, watch all of our episodes because we love you as part of our community. But for this episode, it’s time for us to say goodbye. We’ll see you all later.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
