(gentle music) >> Okay, we're back with Mark Hill who's the director of IT operations at Digital River. Mark, Welcome to "The Cube." Good to see you. >> Thanks for having me. I really appreciate it. >> Hey, tell us a little bit more about Digital River, people know you as a payment platform. >> You've got marketing expertise. >> Yeah. >> How do you differentiate from other e-commerce platforms? >> Well, I don't think people realize it, but Digital River was founded about 27 years ago primarily as a one-stop shop for e-commerce, right? And so we offered site development, hosting, order management, fraud, expert controls, tax, physical and digital fulfillment as well as multilingual customer service, advanced reporting and email marketing campaigns, right? So it was really just kind of a broad base for e-commerce. People could just go there. Didn't have to worry about anything. What we found over time as e-commerce has matured, we've really pivoted to a more focused API offering specializing in just a global seller services. And to us that means payment, fraud, tax and compliance management. So our global footprint allows companies to outsource that risk management and expand their markets internationally very quickly and with the low cost of entry. >> Yeah, it's an awesome business. And, you know, to your point, you were founded way before there was such a thing as the modern cloud, and yet you're a cloud native business. >> Yeah. >> Which I think talks to the fact that incumbents can evolve, they can reinvent themselves from a technology perspective. I wonder if you could first paint a picture of how you use the cloud, you use AWS, you know, I'm sure you got S3 in there. Maybe we could talk about that a little bit. >> Yeah, exactly. So when I think of a cloud native business, you kind of go back to the history. Well, 27 years ago, there wasn't a cloud, right? There wasn't any public infrastructure. We basically started our own data center up in a warehouse. And so over our history, we've managed our own infrastructure and co-located data centers over time through acquisitions and just how things works, you know, those are over 10 data centers globally for us. For us it was expensive, well from a software, hardware perspective, as well as, you know, getting the operational teams and expertise up to speed too. And it was really difficult to maintain and ultimately not core to our business, right? Nowhere in our mission statement does it say that our goal is to manage data centers. (laughing) So, about five years ago we started the journey from our host into AWS. It was a hundred percent lift and shift plan and we were able to complete that migration a little over two years, right? Amazon really just fit for us, it was a natural, a natural place for us to land in and they made it really easy here for us to, not to say it wasn't difficult, but once in the public cloud, we really adopted a cloud first vision, meaning that we'll not only consume their infrastructure as the service, but we'll also purposely evaluate and migrate to software as a service. So, I come from a database background. So an example would be migrating from self deployed and manage relational databases over to AWS RDS, relational database service. You know, you're able to utilize the backups, the standby and the patching tools auto magically, you know, with a click of a button. And that's pretty cool. And so we moved away from the time consuming operational task and really put our resources into revenue and generating the products, you know, like pivoting to an API offering. I always like to say that we stopped being busy and started being productive. (laughing) >> I love that. >> And that's really what the cloud has done for us. >> Is that what you mean by cloud native? I mean, being able to take advantage of those primitives and native API. So what does that mean for your business? >> Yeah, exactly. I think, well, the first step for us was just to consume the infrastructure, right? But now we're looking at targeted services that they have in there too. So, you know, we have our data stream of services. So log analytics, for example, we used to put it locally on the machine. Now we're just dumping into an S3 bucket the way you're using Kinesis to consume that data and put it in elastic and go from there. And none of the services are managed by Digital River. We're just realizing the capabilities that AWS has there too. >> And as an e-commerce player, retail company, were you ever concerned about moving to AWS as a possible competitor, or did you look at other clouds? What can you tell us about that? >> Yeah, and so, I think e-commerce is really mature, right? And so we got squeezed out by the Amazons of the world. It's just not something that we were doing, but we had really a good area of expertise with our global seller services. So we evaluated Microsoft, we evaluated AWS as well as Google and, you know, back when we did that, Microsoft was Windows-based. Google was just coming into the picture, really didn't fit for what we're doing, but Amazon was just a natural fit. So, we made a business decision, right? It was financially really the best decision for us. And so we didn't really put our feelings into it, right? We just had to move forward and it's better than where we're at and we've been delighted actually. >> Yeah, makes sense, best cloud, the best tech. >> Yeah. >> You know, I want to talk about Chaos Search. A lot of people describe it as a data lake for log analytics. Do you agree with that? You know, what does that even mean? >> Yeah, well, from our perspective because the self-managed solutions are costly and difficult to maintain. You know, we had older versions of self deployed using Splunk, other things like that too. So over time, we made a conscious decision to limit our data retention in generally seven days. But in a lot of cases, it was zero. We just couldn't consume that log data because of the cost, intimidating in itself, because of this limit, you know, we've lost important data points, use for incident triage problem management, trending and other things too. So, Chaos Search has offered us a manageable and cost-effective opportunity to store months or even years of data that we can use for operations as well as trending automation. And really the big thing that we're pushing into is in the event of an architecture so that we can proactively manage our services. >> Yeah, you mentioned elastic. So I know I've talked to people who use the Elk Stack. They say, yes, this is exponential growth in the amount of data. So you have to cut it off at whatever. I think you said seven days, >> Yeah. >> Or less, you're saying you're not finding that with Chaos Search? >> Yeah, yeah, exactly. And that was one of the huge benefits here too. So, you know, we we're losing out if there was, you know, a lower priority incident for example and people didn't get to it until eight, nine days later. Well, all the bread crumbs are gone. So it was really just kind of a best guess or the incident really wasn't resolved. We didn't find a root cause. >> Yeah, like my video camera's down you know, by your other house, is that when somebody breaks in, I don't find out for two weeks and then the video's gone, kind of like same thing. >> Yeah. >> So, how do you, can you give us some more detail on how you use your data lake and Chaos Search specifically? >> Yeah, yeah. Yep and so there's many different areas, but what we found is we were able to easily consolidate data from multiple regions into a single pane of glass to our customers. So internal and externally, you know, it really does serve that operational support for the data extract transformation load process, right? It offered us also a seamless transition for the users who were familiar with elastic search, right? It wasn't difficult to move over. And so all these are a lot of selling points benefits. And so now that we have all this data that we're able to capture and utilize, it gives us an opportunity to use machine learning, predictive analysis. And like I said, you know, driving to an event driven architecture. >> Okay. >> So that's really what is offered and it's been a huge benefit. >> So you're saying you can speak the language of elastic. You don't have to move the data out of an S3 bucket and you can scale more easily. Is that right? >> Yeah, yeah, absolutely. And it is so for us just because running in multiple regions to drive more high availability, having that data available from multiple regions in a single pane of glass or a single way to utilize it is a huge benefit as well, just to, you know, not to mention actually having the data. >> What was the initial catalyst to sort of rethink what you were doing with log analytics? Was it cost, was it flexibility scale? >> There was, I think all of those went into it. One of the main drivers, so last year we had a huge project, so we have our Elk Stack and it's probably from a decade ago, right? And, you know, a version point or two or something, you know, anyways, it's very old and we went through a whole project to get that upgraded and migrated over. And it was just, we found it impossible internally to do, right? And so this was a method for us to get out of that business, to get rid of the security risks and support risk and have a way for people to easily migrate over. And it was just a nightmare here consolidating the data across regions. And so that was a huge thing. But yeah, it has also been the cost, right? We're finding that cheaper to use Chaos Search and have more data available versus what we were doing currently in AWS. >> Got it, I wonder if you could share maybe any stories that you have or examples that underscore the impact that this approach to analytics, >> Yeah >> Is having on your business, maybe your team's everyday activities, any metrics you can provide, >> Yeah. >> Or even just anecdotal information? >> Yeah, yeah. And and I think, you know, one, coming from an Oracle background here, so Digital River historically has been an Oracle shop, right? And we've been developing a reporting and analytics environment on Oracle and that's complicated and expensive, right? We had to use advanced features in Oracle like partitioning materialized views and bringing other supporting software like Informatic, Hyperion, Essbase, right? And all of these require a large team with a wide set of expertise into the separate focus areas, right? And the amount of data that we were pushing at the KF search would simply have overwhelmed this legacy method for data analysis than a relational database, right? In that dimension, the human toll of the stress of supporting that Oracle environment than a 24 by seven by 365 environment, you know, which requires literal or no downtime. So just that alone, it was a huge thing. So, it's allowed us to break away from Oracle, it's allowed us to use new technologies that make sense to solve business solutions. >> You know, Chaos Search is just a really interesting company to me, I'm sure like me, you see a lot of startups. I'm sure they're knocking on your door every day. And I always like to say, "Okay, where are they going after? "Are they going after a big market? "How are they getting product market fit?" And it seems like Chaos Search has really looked that hard at log analytics and sort of maybe disrupting the Elk Stack. But I see, you know, other potential use cases, you know, beyond analyzing logs. I wonder if you agree, are there other use cases that you see in your future? >> Yeah, exactly. So, I think there's one area would be Splunk for example. We have that here too. So we use Splunk versus, you know, flat file analysis or other ways to capture that data just because from a PCI perspective, it needs to be secured for our compliance and certification, right? So Chaos Search allows us to do that. There's different types of authentication, really a hodgepodge of authentication that we used in our old environment, but Chaos Search has a more easily usable one, one that we could set up, one that can really segregate the data and allows to satisfy our PCR requirements too. But Splunk, I think really, deprecating all of our elastic search environments are homegrown ones, but then also taking a hard look at what we're doing with relational databases, right? 27 years ago, there was only relational databases, Oracle and SQL server. So we've been logging into those types of databases and that's not cost-effective, it's not supportable. And so really getting away from that and putting the data where it belongs and that is easily accessible in a secure environment and allowing us to push our business forward. >> And when you say where the data belongs, it sounds like you're putting it in the bit bucket S3, leaving it there, >> Yeah. >> And this is the most cost-effective way to do it and then sort of adding value on top of it. That's what's interesting about Chaos Search to me. >> Yeah, exactly, yup, yup versus the high price storage, you know, that you have to use for a relational database, you know, and not to mention the standbys, the backups. So, you know, you're duplicating, triplicating all this data in here too in expensive manner. So yeah. >> Yeah, copy creating, moving data around and it gets expensive. It's funny when you say about databases, it's true. But database used to be such a boring market now it's exploded. Then you had the whole no SQL movement and SQL became the killer app, you know, it's like full circle. (laughing) >> Yeah, yeah, exactly. >> Well, anyway, good stuff Mark, really, I really appreciate you coming on "The Cube" and sharing your perspectives. We'd love to have you back in the future. >> Oh yeah, yeah, no problem. Thanks for having me. I really appreciate it. >> Yeah, our pleasure. Okay, in a moment, I'll have some closing thoughts on getting more value out of your growing data lakes. You're watching "The Cube," you're leader in high-tech coverage. (gentle music)