PurePerformance - Re-Engineering Autodesk: Digital Transformation with AWS Lambda with Daniel Khan

Episode Date: January 29, 2019

Daniel Khan talks about how Autodesk transformed customer’s businesses using AWS Lambda by simplifying processes, demonstrating real-world examples about how Dynatrace provides deep insights into Lambda functions and efficiency.

Coming to you from Dynatrace Perform in Las Vegas, it's Pure Performance! Hello from Dynatrace Perform 2019 in Las Vegas. I'm Andy Grabner, and this is Up Close and Personal with, well, former product management, now Innovation Lab again, on Pure Performance. I want to introduce my guest, Daniel Kahn. Daniel, awesome that we are both here in Vegas. Awesome, yes. Yeah, I just said, you know,

I'll do these sessions with a lot of product managers, and you kind of switched sides from the Innovation Lab to product management and now back to the innovation lab. But you just did a breakout here at Perform, I believe with Autodesk? Yeah with Autodesk. Actually for serverless Lambda I'm still doing the product management because it's still in early access and I was owning this topic for a while and I'm now finishing it until it's generally available. Yeah, that was an awesome session with Autodesk and AWS.

So should I tell you a little bit about it? Yeah, I think so. The idea is, you know, unfortunately, not everybody could make it to perform. So hopefully a lot of people will listen in and we want to tell them what they miss and what they actually get to see and learn in case they watch the recording eventually.

So what were the highlights? What did you talk about? Yeah, first, Mike Deck of AWS, he's a solution architect, gave a brief overview about what Lambda actually is or serverless is because this term serverless is meanwhile here or it's becoming a buzzword he said this himself and he kind of framed this whole topic a little bit more so that you know what to expect when you say serverless and serverless is of course when it when we say serverless, it's still servers. And still, in this case, with Lambda, AWS just takes care of the whole provisioning,

autoscaling, and all of that. So he described very well what this is all about and how this works. Yeah, and then it was Sami and Josh of Autodesk that were talking about their very specific or their specific use case. I don't know if you know Autodesk. I know the name, right? I'm sure I have some idea what they did in the past, but I guess I don't really know

what they do these days, especially with Lambdas or serverless. I mean, Autodesk, first of all, is of course a software-first company. It's a software company, and they come from desktop software. And everything from AutoCAD, which like every engineering company uses but also rendering a software I guess it is Maya that is by Autodesk that you see in each and every movie this whole 3d rendering stuff that's also Autodesk so Autodesk is very powerful so you without knowing you come in touch with their work in the one or the other way constantly. And what they did is they had first a monolith to take care of

all this software licensing because Autodesk as well as every vendor nowadays does, changes their way to sell software to a subscription-based model. And they had a monolith for that and then they transformed everything around that to using Lambda, to using serverless. And it's basically the most used part of their of the public facing web offerings is this lambda part because that's the part where you sign in the cetera where all all this this licensing licensing stuff is going on and there they yeah Josh talked about how how what they did basically so why they did it, the whole reasoning about why. That's, I think, an interesting point. Why should you go to Lambda or to serverless?

I would assume that's probably then. So from monolith to serverless, I would assume a rewrite, or how did that work? Did they take individual pieces first and then gradually kind of replace, like the strangler pattern, strangle things out or how did he do it he mentioned that um i cannot say in detail from the session but he mentioned that if you want to like shift large parts of stuff you might be better off with cost with containers but regardless of that you can do everything with serverless, but it may be a little bit more effort to move everything into function. Sometimes you have to always be pragmatic.

It's easy, like if the Defender story is so famous in this, when you build something from scratch, you can basically go all serverless. But if you do monolith to microservice migration, you have to do it step by step, I would say. That was what he was saying. Pretty cool. So what else? From a monitoring perspective, obviously they use Dynatrace, I would assume, right? Because otherwise they wouldn't talk at our conference. What are their lessons learned, best practices? They are just rolling out Dynatrace and they have, like Sami described,

which key performance metrics they care about, what is important for them, and how they use now Dynatrace to monitor their Lambda function. Of course, everything that applies to regular monitoring applies here as well. You want to have insights into response times of function. Of course, you want to see into errors. You want to see with with

node.js everything that is going on with memory that is going on because also especially with lambda because in dynatrace you see how much memory your lambdas are actually consuming and how much CPU they are consuming and this gives you also ways of course to optimize your lambda. So that means you can do right sizing? Yeah, you can, yeah, yeah, exactly. Right sizing is the word, awesome, yeah. Yeah, you can do right sizing there, yeah. So that's an important point. You can, with Dynatrace you have code level visibility and all of that. So all of

that helps them to better understand what's going on because one thing, and that's maybe the big drawback or the problem that is always mentioned with Lambda is debugging or finding errors in Lambdas is hard. You have to go through a lot of CloudWatch logs and figure out what's actually going on and there also Dynatrace helps you. That's pretty cool. Are they using, because that's a topic that is dear to my heart, are they using Dynatrace only that is dear to my heart, are they using Dynatrace only for production monitoring of the Lambda functions? Although, do they have like a pre-prod, a blue-green environment that they also monitor? Or how does that work?

They did not talk about the DevOps chain in this case. I cannot tell. They are just, of course, in pre-prod, they have it. They are testing testing with it but if they have it like along the line of their whole pipeline I could not tell cool Daniel anything else any other highlights anything else you want to tell people either about the session or something that you would like as you know still being responsible from a product management perspective things that people should know that listen to this, that want to go towards serverless, any resources, anything you can point out? Right now I have this one blog post about how to deploy Dynatrace with serverless. We also meanwhile have for the serverless framework is the most

popular way to deploy lambdas and serverless function at all. And we meanwhile have even a module for Dynatrace dedicated to the serverless framework that's really, really easy to get Dynatrace in there and deploy it. So it's really simple. And I can tell that we are right now also exploring to get this whole process even simpler with new capabilities AWS gives us, but cannot talk about that yet.

Also this is still in EAP, so it's an early access program. You find the link also on my blog post. If you want to try it out, then just sign up for the EAP and you will then edit and get a notification a few days later. And I'm also like, especially in this stage of product development, we are also relying on feedback. So I'm happy if someone enters an EAP, I would also appreciate if they get back to me with what they found out. That's perfect. Cool. And for those listeners that actually happen to be in Vegas right now,

I'm sure they can track you down in the exhibition area in the innovation lab, right? Exactly, yeah. We have our towers there in this innovation area there. And you can find me. Perfect. Cool. All right. Hey, thanks, Daniel. Thanks, Andy. For Pure Performance, I'm Andy Grabner. Bye. Cool. All right. Hey, thanks, Daniel. Thanks, Andy.

For Pure Performance, I'm Andy Grabner. Bye-bye.