 It's a difficult act to follow, like standing between you and your lunch. So I'll try and finish it before the time, so that you can have a nice lunch. I'm really, really excited to be here. This is my first trip to Singapore. And it's been like, it's been like, I've come here yesterday morning and it's been a nice time from then on. I'm going to be talking about the evolution of serverless, past, present, and future. When are you trying to predict the future? It's always a very, very tricky thing. A lot of people got their fingers burned trying to predict the future, especially with respect to technology. Let's see how I do. So what we'll be doing is first try and discuss a little bit about what is serverless. So there are a lot of misconceptions when it comes to serverless. So we'll talk about them. And I'll take a brief peek into the past and see where we are today. And also we'll be looking at what my predictions for the future are. And what would be the takeaways? Yeah, I think one of the takeaways would be the fact that when it comes to technology, a lot of things don't have a clear definition. What does DevOps mean? I mean, there's not a single definition. If you give an example, you can think of whether it's DevOps or not. But what is DevOps itself does not have a precise definition as such. And that's true for a lot of things with technology. And I think that's one of the things which is true with serverless as well. Other than that, one of the things I observe is that the adoption of serverless is increasing. So that's another thing that we would discuss. Another thing I really believe in is the fact that almost all cloud services and almost all architectures are heading towards serverless. So that's another thing we'd be talking about. So we'll discuss these in addition to the other serverless trends I'm observing. Just a big background about me. I'm Ranga. I'm founder of In 28 Minutes. I'm a very popular Udemy instructor. I have about a million learners on Udemy who are doing our courses on cloud DevOps programming. And yeah, you can connect with me on rangacarnum.com. Let's start with a few kind of jokes about serverless. So why did the serverless function go to the beach? Any guesses? Because it's always running in the cloud. It needs some sun. Why did developers switch to serverless? Yeah, this is actually true. A lot of us are very, very tired of playing with servers. You don't really want to do manage softwares, installations, your hardware. You don't really want to worry about that. Why was the serverless function feeling cold? Any guesses? I think it should be easy. Yeah, because it was running on a cold start. I think this is something where AWS is putting a lot of effort into making it better. One of the recent features which is added in SnapStat, I think that really makes a difference when it comes to Java, cold stats. But I think there is a long way to go with respect to cold stats. Yeah, how do you make a serverless function feel better? Any guesses? Obviously, give it more hardware. So give it more CPU. Give it more memory. Yeah, just off. So whenever we talk about serverless, there are a lot of misconceptions. The word itself, serverless, it almost says that there are no servers. But when it comes to the ground, your applications are running somewhere. Your applications are finally running on servers, whether you're making use of serverless or not. So that's one of the most important misconceptions about serverless. When somebody is starting with serverless, they would be thinking, hey, where are my applications going to run? So serverless, it's not about servers not being there. But it's something a little different. So and another misconception I very, very frequently see is that a lot of people think we are past the wave of serverless. So if you look at the entire journey of serverless around the time frame when lambda was brought in, that's when the word serverless became a little bit more pronounced, more popular. And I think within a few years, like the 2020 time frame, 2018, 2019, 2020, there was a lot of hype around serverless. Everybody was saying everything will be serverless, and so on and so forth. And none of them has actually materialized in the last few years. And now the buzz around serverless, it's not so much. Today, hardly anybody talks about serverless being something really, really big. So that's another misconception, the fact that the amount of use of serverless is going down. So that's another misconception. Actually, in my view, as we see a little later, the amount of use of serverless is actually going up. A good case in point was Donnie's and Steve's talk earlier. Almost 80% of the services which were talked about were serverless. You had no instances. You can directly go in and play with them. You don't really need to create a server or an instance or any of that kind. So that's another thing. So another misconception I usually see is usually serverless is actually linked only with compute, right? So a lot of us think AWS Lambda is the only serverless service or AWS Lambda with API gateway is a serverless service. But actually, when it comes to serverless, there are serverless databases, data warehouses, right? So services like SQS, SNS. So there are a lot of other things that you can do in a serverless approach as well. And another very important misconception I see is serverless is always cheaper, right? That's not actually true. If you don't get your architecture right, then serverless can be much more expensive than going the server way. So that's another thing which we have to be really, really careful about. So let's get to the first one, right? The first misconception about what is serverless, right? So whenever we want to deploy an application, typically if you go back a few years, what we would be thinking about is what kind of server I would want, what kind of hardware I would want for that specific server, what do I want to install on that specific server, right? So those are the important decisions we think about, how do you scale the application up? How do you make it available, right? So those are some of the factors that we think about before we even think about writing some code or thinking about the use case or things like that. So we have to make a lot of decisions before you get into the business logic and your code. So for me, serverless is just a trend towards taking your focus, putting your focus into your code and your business logic rather than focusing on the infrastructure, right? So it's all about trying to focus on your code, on your data, on your service, on the business logic that you're trying to implement, rather than focusing on where it would run and what is infrastructure you'd need, how to make it highly available, how to make it highly scalable. So it's about shifting away some of the response, like AWS uses this term a lot, undifferentiated heavy lifting, right? So if you are an insurance company or if you are a health company, right? So you'd want to focus on your core business logic. You don't really want to focus on infrastructure. And for me, serverless is just a way where you can take the next steps, right? So you would want to focus on as much as possible on your business logic rather than focusing on other things. And you'd want to leave things like scaling, things like availability. You'd want them to be the headache of the cloud provider rather than they become your headache, right? So that's kind of how I see serverless. There are a lot of definitions out there. For me, serverless is just about giving as much responsibility as possible to the cloud provider. You don't want to worry about a lot of the infrastructure details. As many details you'd want to leave it to the cloud provider. That's like, for me also serverless is not about, I mean, it's not whether you're doing serverless or not. It's at the level, right? So there is a serverless scale starting from zero to 100. It's about where you are on that specific scale. There is no 100% serverless at all. So it's about how much responsibility you are giving away to the cloud provider that's where you'd be grading yourself. Probably, in my view, it's like, you might be 80% serverless or 90% serverless, but it's not an absolute thing. It's not a one or zero thing. It's like, when you're designing your architecture, it's like, depending on how much you're worrying about the servers, there's a grade. I kind of think of a fit like that. So for me, actually serverless is kind of an operational model where you would actually, like, you'd not worry about servers and one of the characteristics of these models is paying for use, right? You're not paying for number of instances. You are actually paying for the number of, I mean, the amount of consumption, amount of memory you're making use of, CPUs you're making use of. So you're paying for consumption rather than the number of instances which are present in there. And I kind of see that reflect in this survey which was done as well. So this is actually a little older report. This is not the most recent one. This is IBM Serverless in the Enterprise 2021 report where they are looking at, like they have done a survey like among users of serverless on what are the benefits that were experienced. So they have implemented like a number of enterprises and implemented serverless and what were the benefits that they have experienced. I think one of the most important benefit is redemption in amount of management effort that you need to put in. I think that's exactly what's reflecting in there as well. The other improvement, the other major improvement you can see in there is the improvement of application quality. So for me, I think the first three, right? So in terms of reducing the amount of management effort, reducing, improving your application quality and also having greater flexibility, right? Scaling, availability, these are very, very important. So having greater flexibility. I think those are the three main advantages when it comes to serverless architectures. So for me, serverless is about reducing the responsibility, reducing your responsibility as much as possible and leaving things out, like leaving things which are undifferentiated, heavy lifting to the cloud provider. It's about focusing on your business logic. It's about focusing on your code or data and it's about paying for use. And how many of you are aware of Gartner hype cycle? Right, a lot of you. So a quick introduction to Gartner hype cycle. Basically any technology kind of goes through this cycle, right? So you have a technology trigger where somebody says serverless, right? They introduce the term serverless and at the start, within the first couple of years, you would have the peak of inflated expectations. Basically, you would think everything will be that, right? So the same was the case with serverless as well. So when serverless was introduced after a couple of years, we were thinking everything will be serverless. Almost all architectures will be serverless, right? So after a little while, what would happen is you'd start realizing the real use cases. You'd think, okay, maybe this is not the right fit in this specific scenario. So you'll get more data. I mean, you would have implementations, you'd get data from that and you'd see where, which are the right places to use a specific technology. And that's where you'd have the trough of disillusionment. So it's basically, after the hype comes the down and slowly we would go towards identifying the exact use cases where that might be the right scenario to use, right? So you'd see almost every technology go through this, right? So serverless, Kubernetes for example, Docker. So all these technologies goes through this hype cycle, right? So for me, I think when it comes to serverless, we are somewhere in between the peak of inflated expectations and the trough of disillusionment, especially because, I don't think the expectations are huge right now. In the last few years, the expectations on serverless have went down. And I think we are somewhere very near here. And I would think within a little while, we would start seeing the real benefits of serverless. And when it comes to serverless, another typical misconception with respect to serverless is serverless is AWS Lambda, right? Right, AWS Lambda is where serverless terminologies started. That's where a lot of us started hearing this word serverless. But actually serverless existed even before AWS Lambda. And also today, serverless is well beyond just compute, right? So you have things like DynamoDB for example, right? So whenever you'd want to make, typically when we talk about databases earlier, you need to create a server, you need to then create instances, then you'd be able to create your data in the server. But when it comes to things like DynamoDB, you can directly go in and create your database, right? So when it comes to, for example, a data warehouse, right? So a good example for serverless data warehouse from the beginning is BigQuery, right? So when it comes to BigQuery, Google Cloud BigQuery, it's like you can store as much data as you'd want and you'd only pay for the data you are storing. So you'd pay for the data you store and you'd be paying for the queries you do of the data. You'd be paying for operations. And you'd see that BigQuery is actually something which is present from 2012, right? So it's like the term serverless was coined from the time AWS Lambda became popular. But for me, the concept of serverless, not worrying about servers at all, focusing on your data, paying for the volume of data, paying for the operations, that existed well before DynamoDB. A good example in AWS is AWS S3, right? So AWS S3 was actually introduced in 2012, right? Even with AWS S3, you don't really create a server. You'd actually create a bucket and you'd pay for whatever you store in the bucket and you'd pay for the operations which are performed on the specific bucket. So yeah, the serverless terminology itself, even though it started with AWS Lambda, in my opinion, cuts across almost all services, right? So you have functional service offerings like AWS Lambda. You have similar services which exist in other services, other cloud platforms as well. You have cloud functions in Google Cloud. You have Azure functions in Azure. And also today, like one big evolution with respect to serverless is the evolution towards containers, right? So when AWS Lambda came in, probably containers were not so big. Docker was not so widely used. But today, if you look at most of these services which are related to containers, for example, ECS in AWS. Now has a serverless version, Fargate, right? So and it's improving every day. Similar to that, almost every other cloud platform also has a serverless way of running containers as well. And like when it comes to databases, you have things like DynamoDB, which is serverless. And when it comes to data warehouses, like Google Cloud, BigQuery, like Azure Synapse Analytics, and also like Redshift now has a lot of serverless features as well. Like one of the biggest challenges when it comes to data warehousing is managing the servers, right? So you have huge volumes of data to store and you want to query from it as well. So yeah, it's a space where serverless makes a huge difference because you only want to pay for the data you're storing and whenever you want to execute the query, you don't want to only pay for the query execution. You don't want to pay for all the servers which are present in there. And that's something which we have a lot of improvement when it comes to AWS with respect to Redshift and the serverless options which are coming in into Redshift. And also like another very, very important area where there's a lot of focus in serverless is application integration, right? So there were a lot of use cases that Donnie and Steve presented to us related to application integration, right? So we looked at the integrating AWS services and we also looked at how to integrate SaaS services with AWS. And one of the key things that we would have observed, it's kind of taken for granted right now is the fact that we are not worrying about, for all the services that we looked at earlier, we will be paying for consumption. Like once we create a flow, you will be paying for the number of executions of that flow or how many operations were performed and things like that. And you'd not be really paying for the servers themselves. So it's also serverless is becoming more underneath the hood nowadays. It's almost taken for granted kind of. So we already looked at the, I mean we already talked about the serverless options for compute, right? So AWS Lambda, AWS Fargate is the serverless compute engine which works with Elastic Container Service. And AWS Fargate can also be used with Elastic Kubernetes Service as well. And we also have a number of data stores which are serverless, right? So you have S3, EFS, DynamoDB. You have RDS Proxy which can actually integrate with RDS. Amazon Aurora is a very great, good example of AWS Arora serverless as well. And you have Amazon Redshift serverless coming in as well. And you have a lot of things. We already talked about a lot of these. Like we talked about EventBridge earlier. Like we talked about the like multiple, like a family of EventBridge. And we have step functions, SQS. All these are good examples of things which help us to implement serverless. And serverless architectures are also implemented in wide variety of use cases, right? So the typical, this is kind of the most popular architecture, right? So for building most popular use case of serverless, I think to build things like web applications, to build things like REST API. So you have your data stored in DynamoDB which is exposed, which is like which a Lambda function would be talking to. And this would be exposed from something like API Gateway or something like that. So this is one of the popular architectures when it comes to REST API with serverless AWS Lambda. The other like architectures which are also possible are data processing architectures, right? So this is an example of actually like you have nodes coming in and you don't want to store them to S3. And you don't want to do further processing based on the nodes. Maybe you'd want to generate a PDF or things like that. So those kind of things can easily be done in an event-driven style as well, right? So you have an object which is stored in S3 and you can have events triggered from S3 and you can trigger Lambda functions or you can put messages in the queue and you can do a lot of processing after that based on whatever is coming in. So you can process the biggest advantage over here is once you create the entire flow, you don't really need to worry about scaling it or you don't need to worry about availability of it and things like that. You can also implement batch programs, batch processing in a serverless way as well. In the previous session, Donny talked about EventBridge, right? So in EventBridge, one of the new features which came in is scheduling, right? So you can schedule like as soon as, so you can schedule something in EventBridge to trigger off your batch and you can process the entire batch in a serverless way as well. Yeah, you have a number of other things that you can do in a serverless way as well. You can trigger off your machine learning workflows or you'd want to do some processing on your images and things like that. You'd want to get intelligence from them. So those kind of things also can be done in a serverless way. So basically the serverless architectures, there are a wide variety of possibilities when it comes to serverless architectures. And let's quickly look at some of the important trends with respect to serverless. So right now, if you look at the serverless market, it's, I would say somewhere, it's like the bottom is completely fragmented and the top is consolidated. I mean like where there are very, very few players. I would say it's somewhere in between, right? So today, like when it comes to function as a service offerings, I would say Lambda has a big advantage. But when you talk about serverless in general, I think it's somewhere in between. And this is from Mordor Intelligence Survey. And you can see that there are a wide variety of use cases where serverless is being used. Again, this is from the IBM serverless in the Enterprise 2021 report. You can see that it's used across different domains. And as usual, like AWS has a lead when it comes to like the serverless adoption. So you can see that almost half the organizations in each cloud have adopted serverless. But this is from the Datadoc Survey, like serverless survey 2022. So you can see that almost 75% of the enterprises which are using AWS have adopted serverless. So like AWS is leading the way in terms of how many of its customers are adopting serverless. This is like, you can write your Lambda functions in obviously multiple languages. And you can see that the two prominent languages which have a big lead are Python and Node.js. I think with this snapshot, with this snap start feature, probably Java would improve. But let's see, let's see how it goes. But right now Python and Node are the leading languages when it comes to writing your Lambda functions. And if you, I mean, whenever we talk about like serverless, it's kind of a flow. So you need to integrate your Lambda function. You need to have triggers for your Lambda function from somewhere. And you can see that like a lot of like invocations, the most popular way to invoke Lambda function is API gateway, right? So today, I mean, one of the recent features which was introduced into AWS Lambda is the HTTP invocation, right? You don't really need an API gateway to invoke a Lambda. But still, you'd see that actually, API gateway is the number one preferred option. The other one is through the queue, SQS. So you have a notification, you have a message coming into the queue and the Lambda function is invoked in response to that. The surprise in here is the event bridge, right? So event bridge is something which is pretty recent, I would say, but still it's doing pretty well. A lot of Lambda functions are getting triggered from event bridge as well. So yeah, and one of the like other trends that we see in the last year, like last couple of years is a lot of enterprises are going container first, right? A lot of enterprises, like earlier it was cloud first, today it's like container first, right? So you'd see that a lot of container related technologies are really loved, right? If you look at Docker, Kubernetes, these are really, really loved by a number of developers who make use of them and that's exactly the trend which is reflecting when it comes to serverless as well. The number of services, number of offerings around containers with respect to serverless in each of the cloud platforms are growing, right? And that's exactly reflecting with respect to the adoption as well. So more than 35% of the container organizations use serverless and the reverse is also true. A number of people who are going serverless are making use of containers as well, especially because today Lambda also supports containers. So yeah, as we talked earlier, like I see that with respect to serverless, we are somewhere in between, between the peak of inflected expectations and tough of dissolutionment and I see the adoption of serverless growing further and a lot of times we are making use of serverless unknowingly. Let's, we have three more minutes. So let's get to the predictions. Yeah, this is a famous prediction by somebody who's very, very popular, much, much more popular than me and he said 640K needs to be sufficient for anybody and this phone in here today has more than 64 gigs of memory, right? So it's like predicting is very, very difficult. Yeah, this is something which we are already observing, right? So almost like I would say serverless would become mainstream and a lot of times you would not even think about serverless as something specific, right? A lot of services you would make use of are serverless underneath and that will be the trend going forward. Probably I think very, very soon, like it's been almost seven, eight years since the term serverless was coined. I think there would be a new name for serverless, right? So in technology we have a way of actually doing the same thing and giving it a new name and I think we will find a new name for serverless very, very soon. Almost all the cloud services would become serverless and yeah, and like with respect to containers, I think like almost everything related to compute and serverless, I feel we would consolidate towards the containers way, right? So I think serverless containers will become the norm and obviously another factor where I think serverless service, I mean serverless architectures and serverless flows can improve is with respect to the developer experience, right? So building serverless is not as easy as building a usual traditional application, especially when it comes to debugging problems, right? So there are a lot of things that we can do better when it comes to that. Each of the cloud providers is doing a lot with respect to that but I think still there's a long way to go with respect to improving the developer experience with respect to building serverless applications. Yeah, now like my recommendations when it comes to like how do you get much more knowledge about serverless? I love this serverless ICY MY, it's in case you missed it. This is something which is published on the AWS website every quarter and I love reading it, right? This gives you a complete overview of what's happening with respect to serverless every quarter and this is a great place to look at if you would want to explore serverless further or if you want to understand the history of serverless, how it evolved. Another great place to go is the AWS well-architected framework. The serverless application lens is a great place to go in and look at all the architectures that are possible with respect to serverless. Serverless land is another website I love. This is where a lot of patterns related to serverless are present. So you can actually see around 400 plus different templates with respect to getting started with your serverless architectures. And I love the Data Docs survey. They have been publishing this for a few years now and this is a great place to understand what's happening in the serverless world. Thanks a lot. I think serverless is here to stay. We might find a new way to express what serverless is, but I think serverless is something which will be here to stay for a long time. Thanks a lot. I would want to thank the organizers, Steve, Donnie and Badri. Thanks a lot. You have helped me a lot in actually even getting a visa and coming in here. Thanks a lot for all that. I hope you all have a nice session and thanks a lot. Thank you.