Episode #33: The Frontlines of Serverless with Yan Cui

About Yan Cui:

Yan is an experienced engineer who has run production workload at scale in AWS for 10 years. He has been an architect and principal engineer with a variety of industries ranging from banking, e-commerce, sports streaming to mobile gaming. He has worked extensively with AWS Lambda in production, and has been helping clients around the world adopt AWS and serverless as an independent consultant. Yan is an AWS Serverless Hero and a regular speaker at user groups and conferences internationally. He is also the author of several serverless courses.

Transcript:

Jeremy: Hi, everyone, I'm Jeremy Daly and you are listening to Serverless Chats. This week I'm chatting with Yan Cui. Hi, Yan. Thanks for joining me.

Yan: Hi, Jeremy. Thanks for having me.

Jeremy: So you are a developer advocate at Lumigo. You are an AWS serverless hero, you are also an independent consultant and I think more people know you as the Burning Monk. But why don't you tell us a little bit yourself and what you've been up to lately?

Yan: Yeah. I'm all those things you just mentioned. I'm doing some work with Lumigo as a developer advocate where I'm focusing a lot on the open-source tooling and articles and in my sole capacity as an independent consultant I also work with a lot of clients directly. A lot of them are based in London where I used to be based. Nowadays I moved to Amsterdam. I still do a lot of open-source work. I just started a new video course focusing on Lambda best practices. Then I'm also doing some workshops around the world. In Europe and also now looking at U.S as well. So doing a lot of different things to keep myself busy.

Jeremy: Awesome. Listen, I can talk to you probably about anything. Anybody who knows or seen some of the work that you've done, it's quite expansive. It's very impressive. In 2019, I have some numbers here, you did 70 blog posts, something like 2200 students to your video courses. You spoke at 31 conferences in 17 cities. But more importantly, you helped 23 clients in 11 different cities. So you are on the front line here in seeing how companies are adopting serverless. And not from one perspective and I think that's what we get a lot from different companies is, there is one perspective of how they adopt serverless and how they are working with that.

You've obviously seen this from multiple perspectives, so just, I want to talk about adoption a little bit. We'll talk about a few other things, but just what are you seeing with companies now? The customers you're working with or the clients you're working with, what are they using serverless for?

Yan: They are using it for all kinds of different things. I think depending on, I guess the maturity of the company, the domain they're working in. I've got a lot of clients that they are either enterprises or a lot of small and medium sized enterprises, and even some stealth-mode to startup as well. And obviously your constraints are completely different. That's one of the things I really enjoy about being a consultant. Where I get to see a lot different perspectives and what may work for one company may be completely inappropriate because the constraints a different company would have. So in terms of the adoption patterns, you see a lot of the, I guess startups that are in that position where they can go all in on serverless.

They are your great serverless-first going to the game. But then at the same time, you also have lots of, I guess midsize companies and enterprises. They have so much existing intellectual properties that it wouldn't make sense for them to rewrite everything just so that they can run code on Lambda. For all those companies, you see a mix of Greenfield projects. They are serverless first and then at the same time there's some effort to migrate some of the existing projects to work on serverless at least to some degree, at least gradually. Of course, depending on a lot of constraints around how much of on-premises stuff you have.

Do you have to run everything in Java in which case it is the cold start performance that's a concern. So a lot of those limitations I guess affects how quickly and how much you are able to go in on this whole serverless first mindset that we like to have and I think that is probably one of the reason that serverless adoption hasn't been as fast and as many people expected a few years ago because, the fact that, you can't just lift and shift your weight anymore. It means that you always have to allow more thought process behind it and planning and also just risk involved if you make a big mistake and it's your flagship product and of course that's going to put you in a really difficult position. But we do see that companies of all sizes and all fields and all industries are adopting serverless for lots of different workloads, not just APIs but a lot of data processing, IoT, you name it.

Jeremy: That's actually one of the things I'm curious of too. You mentioned customers in all different industries, which is really interesting. Because we get to this point now where I think every company is a software company. Everybody is building some sort of software now. But so, what are the constraints that these companies are working in?

Yan: A lot of them, I guess again, it depends on the industry you're in. For finance companies you have to be very careful about a lot of the, I guess, regulated requirements. In terms of how you handle data and also in some cases you having a plan in case you have to move away from a database for example, that's where some of your vendor lock-in arguments start to kick in. And also for example, you have enterprises who have millions lines of Java code that has been accumulated over 10 years. It's not possible for them to move everything into Lambda if they're seeing one to three seconds of cold start time on those user facing APIs. So some of those constraints are being lifted.

At least they are now getting better with new features on the platform but still it's something that people have to be aware of and also have to understand the mitigation strategies, which a lot of times is where the constraint is, is a lack of knowledge and knowhow because you can even think of Lambda as the extension to a lot of AWS offers, then it means that, you can't just know is it to visualization, you have to know a lot of different services to take full advantage of serverless.

That's where I think a lot of companies are struggling, is that they just don't have the skillset available in-house. They're exposing developers to things that they've never had to think about before. And I think that's where you get a lot benefit from serverless from having autonomous teams that can be self sufficient and look after so many different things, but at the same time, a lot of developers are just not used to working that way. They're used to working in silos where they have very few responsibility, just write your code, someone else will manage running the code in production. They'll manage the infrastructure, but now more of that is your responsibility which can be a gift, but it can be a challenge to companies that are not used to working that way as well.

Jeremy: Yeah, I totally agree. And I think that, as you mentioned learning all these other services, I think we're at a point now where most of these use cases there is some sort of ...