 from the Regency Center in San Francisco. It's theCUBE, covering Serverless Conf San Francisco 2018, brought to you by SiliconANGLE Media. Hi, I'm Stu Miniman, and this is the CUBE's coverage of Serverless Conf 2018. Happy to welcome back multiple CUBE alum, Erica Windish, who's the CTO and founder of IO Pipes. Great to see you. Thank you. All right, so Erica, I talked to you last year at AWS New York City. You also talked to John Furrier and the crew at AWS New York City this year, but it's the first time we've had you on at Serverless Conf. So, before we get into some of the technology, about 500 people here, you're speaking at the show, you go to a few shows. Tell us what you think about the community and the show. I mean, so it's the first day, for me at least, because we have workshops and a hackathon, but this is the first day of the proper conference. And for me, I have to still kind of feel about the audience and see who's present. Honestly, I was a little afraid that these shows might've been turning more into more vendor-oriented, which I don't think is necessarily the case. I think it's been good for Serverless Conf to be in strong developer centers because only two, three times a year, they're doing more, but Serverless Days, which is another conference, is more like DevOps Days. It's like, any city that wants to host it, does it. And this conference is a little bit focused on being bigger, but then that means it has to be focused on Hubs, like San Francisco, New York, Austin, et cetera. And I think it looks like it's doing a good job at that, but again, it's the first day, so I've only been able to talk to so many people. Yeah, absolutely, and I just spoke to Michael Garski from Fender, so they do a good job reaching out to some of the users. And what I've found, at least, there's many of the vendors that are here that they're using Serverless, they're developing Serverless, so companies like yours. So, give us the update on IOPypes. Sure, so IOPype, we've been around for a little over two years now. Let's say, we're selling, we have a product that is useful and usable, and I would say pretty critical to building these next generation of applications, because for me, the products evolve so much, probably since the last time we spoke, I mean, definitely since the last time we spoke, but also just in the last two years, we started out of, okay, how do you understand the Serverless applications, but people we were talking to were thinking about us and looking at us and comparing us to infrastructure monitoring. And I see our story changing to more about application visibility and application insights. Like, how are users using your application? When your application, let's say your database fails, like your computer server's not going to die anymore, like Lambda's going to keep working. Well, when your database goes down, what do your application do? Like you, let's say you're looking for a user and user didn't exist to try and then create a new user. Like, what's your failure mode, right? Robustness of software is critical, but what happens a lot of times when your software doesn't work, you don't know why, or rather, you might know why it didn't work, but you don't know what that caused. You don't know what the unintentional side effects were, you know, those unknown unknowns. And, you know, tools like IOPype are really useful for like, okay, I can now track a conversation for an Alexa scale and see how the user interacted with the scale. And if something does go wrong, I know exactly where it went wrong and what we told the user as a response, right? So we can improve the user experience, the developer experience, you know, we have tools for like, things like understanding what code path your application went down, and if it's going down the wrong path or what percentage of times it's going down, you know, an unusual code path, for instance. Yeah, so Erica, you said a lot has changed since we spoke a little over a year ago. Most people, when they hear about a technology and they hear about, oh, here's what it is and here's what it is, it kind of fossilizes in their brain. Sure. This space is changing so fast. Gives a little insight as to what you've seen, you know, serverless in general and how that's impacting your product. Well, I think serverless is definitely being used for more things, you know, a lot of the early use cases are operations and it's a wonderful use case for serverless. It's a great way of getting used to it and becoming, you know, accustomed with the development model. But you know, we are seeing larger applications moving to serverless. We're seeing more lift and shift of applications of serverless than we did before. Is that a good thing? I've talked to a number of people in the community and they're like, this is wrong, don't do it. I would see this. I think, so the term like, you know, there's pushback on the term serverless. Okay, yes, there's still servers but also state, stateless applications. There's still state. And if you stateless, or I'm sorry, serverless applications or serverless infrastructures like Lambda is a stateless infrastructure for stateless applications. If you can describe your application as stateless, it is a good match for moving to serverless. You know, it's one of those things like if you have a highly stateful application, then no, you're not going to have an easy time migrating. But if you have stateless applications, a stateless architecture, if you've already been, you know, going down that stateless route hall, then serverless may not be as hard of a transition as you think it is. Excellent, that's great clarification. I appreciate that. What else, what are you finding, what advice would you give to people out there that are fogging their way through? Well, I mean, you know, try it small, right? You know, have little hack-day projects, have little operations projects for place cron jobs. You know, serverless can save you a lot of money but like honestly, start with like reducing things that take up time from your ops team, right? If your ops team is spending time on this thing, then put it in the serverless function. Yeah, that's great advice because, you know, our human resources are one of the most expensive things and if you've got good people, they can focus on more business value to the business. All right, last question I have for you, give us kind of the roadmap and outlook for IOPype, you know, what should we be looking for kind of for the next 12 months? So one of the features we're going to be announcing very shortly, I guess a little pre-announcement, we're doing what we call auto tracing. So we'll be able to auto inspect HEP calls that originate from your functions. So you'll be able to have automatic inspection of things like you put a file in the S3. Okay, well, what file did you put in the S3? You know, how big was the file that you put in the S3? What was its object name? You make other API calls, you'll be able to see those and trace them, their timing, their parameters, things like that, their headers, automatically out of the product, which of course you can also filter and make sure nothing that you don't want to go into, you know, into our services don't go out. But you'll automatically get that capturing. Well, some of the things we didn't choose for the last year were things like event inspection. So events come into your function, we automatically inspect them. Now we want to inspect the outcome of your function. So what is your, what's coming into your function and what is it doing, right? That's the IO for the IO pipe. All right, Erica, anything from serverless in general? I guess two things I want to add. One is, or most people you're talking to is still kind of Lambda, AWS centric, and any commentary around kind of multi-cloud and some of the other options out there for serverless and anything you'd be looking for the community to mature, to help this be more useful in the future. So I mean, I definitely hear more and more Azure. I mean, if we're talking about non-AWS, I am hearing more Azure. One of the issues with Azure is they have like kind of like this V2, like next generation of Azure functions, which I'm kind of impatiently waiting for. Because I do think they need, there's a lot of architectural changes that I think are kind of really kind of necessary for that product and they're coming. We've just got to wait for them. And I mean, I think it's just, it's maturity, right? Lambda is deeply mature at this point, or like it's definitely mature enough. And the other services are getting there and they're getting there, they're going to be there soon, but probably waiting another, probably for the next serverless conf to start hearing like, yes, we've been looking at and we're using it. Some of the use cases that work well on Lambda just don't work well, like things like Slackbots don't work as well on some of the other platforms because they're cold start times and they're been packing algorithms, just aren't as good yet, but they're going to get there. All right, well, Erica, pleasure to catch up as always. Thank you. I hope I end everything happening in this space and thank you so much for watching theCUBE. I'm Stu Miniman.