>> Announcer: Live from Manhattan, it's the CUBE. Covering AWS Summit, New York City, 2017. Brought to you by Amazon web services. >> And we are live here at AWS Summit here at the Javits Center, New York City, we're midtown, Manhattan, a lot of activity going on outside, you can imagine all the buzz inside as well. Somewhere between 6, 7, 8,000 attendees, kind of tough to tell right now, but everybody's jammed inside here on the show floor and they've been here all day and they're going to stay for a while I think too. As I said, a lot of buzz going on, and good buzz too. Along with Stu Miniman, I'm John Walls and we're now joined by Erica Windisch who is the Co-founder and the CTO of IOpipe. Erica, thanks for being with us here on the CUBE. >> Thank you, thank you for having me. >> You have had a big day. >> Yes we have, yeah. >> It's always fun to talk about money but you did have a fairly significant announcement this morning to make. Tell us about that. >> Yeah, so this morning we announced funding, $2.5 million from several investors including NEA, Madrona, and Underscore. >> So, yeah, you don't often get to high-five everybody for a day like that. I mean that kind of validation, obviously is something that not you just take to the bank, you take it to the marketplace too. >> Yeah, absolutely. And we actually started, our first check was from Techstars so we joined Techstars here in New York City and did that last year for their summer program and it was really great and that was the first foundation that we really had, and then having that further validation from major VCs like NEA and Madrona, Underscore, you know that really was really validating for us as well as just the fact that we're building, we're hiring and we're building and having what I think is an increasingly awesome product. >> Sure, well tell us about IOpipe, for folks at home who are watching might not be familiar with your space, what you do and how you do it. >> Yeah, so we provide tools for software developers to build and manage their applications on Amazon Lambda. So, basically, it's all serverless, we're actually built on serverless as well, we monitor with IOpipe, we dogfood everything. And we are providing deeper insights into those application workloads as well as correlating that information in more useful ways. Deeper knowledge of what exactly is happening in the run times, so we're able to see the data we ingest tells us information on the processes and the containers and the virtual machines that are running your Lambda workload, so we can see things like memory leaks and we can see file descriptor leaks and displaced utilization leaks, things like that that Amazon doesn't collect or at least doesn't give you that information. So, we're looking at ways we can provide more value to users of Lambda and also extending it with plugins so we have a plugin for tracing where you can time aspects of your application as well as profiler, so you can enable a profiling plugin and you get a full flame graph. So you can see, these are all the functions and this one ran and this one ran and the stack looks like this and so you can see the full flame graph of what happened and when and full timing information. This kind of insight that nothing else really gives you. >> Yeah, Erica, every time we have a new technology we go through this kind of diffusion of innovation that goes through. Remember back, I go back thinking about when virtualization came, people, what is it, how do I use it? We saw that in containers and each wave seems to be going faster and faster so there's still plenty of people I talked to that were like, "serverless what?" You know, some new as a service, I mean I thought I knew it with SAS and everything else like that. You're digging into these environments further. Can you give us, what are some of the kind of key use cases you're seeing, what are the challenges that customers are having? What works, what doesn't work, help us unpack that some? >> So, I think there's a number of challenges that users run into today. One is the fact that it is new so some of the tools are still evolving. Operations tools, development tools are still evolving. Just this week, Amazon announced SAM local so you can do editing and debugging locally on your machine or your laptop. That wasn't available before, right? So these tools, we're very much still in a learning phase for some of the tools, but some of the things like what we're doing with IOpipe, in some ways is more traditional because we're bringing in some of the basic monitoring tools and capabilities that you would expect from other platforms. But the other side, also innovating because we're bridging that development and operations into a single tool so it's not development and operations, it's, not even just different tools for those two things, but single tools for those. So I think that's part of the solution, part of the problem, in terms of workloads, I think there's a lot of ETLs, streaming applications, very infrequent things like chron jobs, web applications, you can take flask applications or express applications and just port them directly over to Lambda with almost a lift and shift for those, right? So there's a lot of power for bringing on the web 'cause you pay per the request. You don't scale your application and build your application for the number of servers that you need to handle the requests, it scales it per request and you pay per request and that's what's powerful in both scale of operations and team and like financially, but also, yeah, I lost train of thought there, but it all scales that way, right? Like just according to the request. >> Yeah, bring us into a typical customer, I know there are no typical customers, everyone's a little bit different, but you've got the developers, you've got the operators, finance has always had, you know, there's challenges with cloud in general but serverless at least promises that it's going to be less expensive. What are those dynamics from an organizational standpoint that you see inside? >> In terms of cost? >> Not just cost, but do the developers make something and the operators are like, wait, you know, there's challenges there? Or who drives this initiative in general? Does finance come and say, has finance heard about this and said hey, I heard I could save 60-70% on my cloud if you just re-architect this on Lambda. Or is it the developers coming through and saying, oh, wow, this is great, and can do it, or are operators, who's driving the initiatives and what are some of those dynamics? >> So I see a combination of these things. Some organizations, and I don't want to say names 'cause I don't want to like, you know, they did this and that's how it is. But I get the impression that certain organizations they have a top-down approach where they're going like, everything is going to be serverless and the cost really matters. So you're going to build serverless unless you can't, right? Serverless by default, anything else as an exception. Then there's organizations where developers are really pushing for it because it simplifies their requirements, right? It's a self-service aspect, right, even if they can spit out VMs, even if they have self-service VMs, they won't have to spit out VMs, they don't have to build docker images, they don't have to look at how the operating system is configured. They write code and they deploy code. There's no other steps, right? They're not like, oh, what version of Python is on here and how do I install all the libraries and how do I, right, like with serverless you just write the code and you ship the code. Which is really, really nice. So, in a way it's like having a golden image that you can't change, and you just know you're always going to build for in every application and every organization is going to the same golden image. Which simplifies a lot of things. >> Stu and I were talking about serverless, the whole concept, because it's really not truly serverless it's just different server, or it's a different flavor of it basically. So, first off, what gave birth to that and then where do you think, with serverless computering, serverless application, so on and so forth, where's that going? >> Yeah. >> What's going to be the real value at the end of the day of that? >> So, first of all the term "serverless," I look at it as, yes there are servers, serverless is servers are not my concern as a developer, right, I am not worrying about what the server looks like or operating the servers necessarily. I care about building my application which is why we're looking at building tools that are bridging development and operations so that operations is part of your development. But I see, the direction of serverless, really interesting in a few ways. One is that it's going to be available for more use cases. So right now there's certain use cases that make sense and one of the challenges is figuring out which use cases it doesn't work for. Eventually, you're not going to have that question, potentially, right? So maybe we get to a point where you don't have to ask, the challenge isn't, is serverless good for this use case? Maybe it's good for all use cases eventually down the road, maybe. Another thing is... >> If I could just follow up on that. Some of the announcements today like AWS Glue has serverless in the background there. Seems very promising, things like machine learning, artificial intelligence, serverless, IOT where you know, I need to balance the surface area of attack there but with serverless it won't be active as much and there will be links that are a little bit more dynamic. So, lots of those new use cases seem to be built really well for serverless. What are some of the cases today that you just say, hey, don't even go serverless there. >> Oh don't go serverless, where to do that? Well, so, Lambda has an execution time window which can be limiting for some things that you might want to do. So, like, Lambda in particular may not be the best case for all video encoding tasks. Some video encoding tasks if you can time limit it can be fine. But it's not good for all video encoded tasks because it's a batch process, potentially. Serverless processes that can let's say paralyze that and say, we're going to run Lambda but we're going to say split this up into segments, for instance, you can do that, or if you do it as a stream, right? Like you pipe a video and blocks into Kinesis, right, you can make that work. But it becomes a challenge to those kinds of use cases. >> Yeah, there was the example I think in the keynote was, this high process that would have taken five years, we can do 155 seconds. >> Right, but you have to paralyze it, right? >> Stu: Exactly. >> And if you can't paralyze a task and you can't do it within five or ten minutes, you can't use Lambda for it today. But it also depends on how you define serverless, right, because if serverless is Lambda, that's one thing. But if serverless is these other SAS products as well potentially, like AWS Transcode service, well is that serverless? If it is, then there you go. There's a solution potentially for you. So there's very blurry lines sometimes around what is serverless, and we're looking at IOpipe around serverless functions. I feel the same way around cloud in general was that there's cloud compute and it kind of evolved over time and the cloud is everything like all these things are in a cloud. But originally when we're talking cloud, five years ago, ten years ago, it was all compute. That's what we were talking about. So these terms change over time, so it's hard to say what serverless will be in five years or ten years because it'll mean something different. >> Or next week, for that matter. >> Yeah. >> Erica, last question I have. $2.5 million, what's that going to drive, what should we expect to see from your company and give us any final thoughts on what you'd like to see for the maturation of the serverless technology field? >> Yeah, so we've been hiring and building out a team, we're working on improving the user experience of the product, we are adding additional plugins and enhancements to the service. We feel that we have a really good base with our 1.0 announcement, 'cause we're not just the 2.5 million, we also announced our 1.0. And the 1.0 has a really good base of functionality and we're looking at adding additional plugins and additional features that can extend the service. So we're looking at doing that with that money. And with serverless in general, I think this is really compelling, what we're going to see in the next year, because we're going to see more large enterprises and more enterprise adoption, I think. I mean I was involved early in cloud. I was involved early in docker. And this point of serverless is very much at the early days of these technologies, and I definitely see a rocket ship taking off, and I think in the next year it's going to be really interesting to kind of see it starting to orbit a little bit. >> Well, new product, new funding, and a new day for IOpipe. >> Yes. >> So congratulations on a good day and thank you for being with us here on the CUBE. >> Thank you very much. >> You bet, we'll continue here at the Javits Center we're in midtown Manhattan continuing our coverage of the AWS Summit, here on the CUBE. (futuristic music)