 Hello and welcome to this CUBE conversation. This is part of our continuing coverage of CUBECon Cloud NativeCon North America 2021. I have a very special guest with us from a technology company that on any given day at any given moment has any number of 31 million discrete users coming in looking for a specific item or two out of 22, 23, 24 million. Who knows how many items that could be shipped from 15, 16, 17,000 different locations around the globe. We've got one of the key folks responsible for managing what some of us in tech would refer to by the technical term a nightmare. Gary White from Wayfair, welcome. Hey, thanks for having me, yeah. So tell us about Wayfair. What is Wayfair? We're all gonna pretend like none of us have heard of Wayfair before despite billions of dollars in advertising. What is Wayfair? What do you guys do before we get into the technology and how it actually works? Yeah, that's a great question. Wayfair exists to be the destination for all things home, helping everybody and create a feeling of being at home. So that's like our primary goal from the overall business objective, specifically in the technology part of the company. We strive to make development of tools that make the shopping process easier. Carrying one of the widest spaces of items means that we have to have incredible resiliency in our data and access to that data. And then we try to build world-class development tools to compete with talent market that is some of the biggest firms in the world. Okay, so just so we're clear, because I wanna make sure you came to the right place. This isn't furniture, Khan. This is a cloud native, Khan. And you're not an artisan crafter of bespoke end tables, right? So you craft a very different thing, which is the technical infrastructure behind this. Tell us about your relationship with cloud native technologies in the open source world. Oh, sure, absolutely. So my experience at Wayfair has mostly consisted of exposing the release engineering process and making the process of developing tools easier. I think most companies invest about, or a lot of companies invest a significant amount of their engineering talent into being able to create a platform for their developers to work on top of. That's the team that I'm a part of where we create a platform for our developers. And a large part of how we do that is leveraging technology that comes out of the CNCF. So we'll be talking about automation tools, things that you may run in Kubernetes to do batch jobs, things that you may run in Kubernetes to run regular microservices and applications, and then ways to automate the building of those applications and the packaging of it so that it can ship to production reliably. And so you guys aren't just, when we go back several decades, there were customers and there were vendors. And those lines are really blurred in the open source community. They have been for a long time, but I mean, you're actually working not only to develop solutions for Wayfair, but you're working hand-in-hand with other people to develop solutions that get propagated across industry. Tell us about some of those projects that you're involved with, with the open source community, or at least where Wayfair is. Absolutely. So Wayfair has made an investment in the open source community, specifically notably with the Tremor project. You can go to tremor.rs and I'm actually speaking about it at KubeCon. That's where the topic comes from. Tremor is a early stage and event processing system where you can give it a lot of data for it to be able to ingest and then spit out downstream to other systems. You may want to send events and notifications out to multiple systems based on what you see. You may want to throttle the amount of data that you have coming in. And that was a big topic that I also mentioned at the CloudNativeCon and at TremorCon where you can deal with this really massive volume that we have to do at the scale that we exist as a business and just filter it so that it doesn't overwhelm everything downstream and observability's sake. Well, you know that Tremor must be cool if it has its own con, right? Yeah, very cool. You know you've arrived when you've had a con. Eventually, I'm working toward DaveCon. We'd make a Dave and Gary con if we need to, but... Oh, absolutely. So can you take those concepts of events and data flow and kind of up-level that in terms of the kinds of things that are happening between customers and Wayfair on a moment-by-moment basis? So give me an example of like, what is an event? Sure, yeah. So events, if you're familiar with the open telemetry or observability framework, you might call them signals where you have something happen on your infrastructure that's processed in a way that you would wanna record. So you might have a log that you need to be able to trace through later in case there's something happens or you just wanna be able to comb your logs. You may have metrics that you're sending in like how many requests you've gotten or how many bytes you've been processing in your service. You may have baggage that you associate with that data. So yeah, all of these different kinds of signals as defined in the open telemetry spec are things that we support in Tremor and we supported before the hotel project really made it a form. It was something we did out of necessity and similar products that you might find are like log stash and elastic and that whole stack. Okay, so if I'm looking for something for my home office, although you can see I've got tons of garbage already in my home office, what happens to me as a customer if all of what you just described goes wrong? What happens to my experience? That's a great question. So as we're talking about Tremor, if something were to go wrong, it shouldn't impact the experience of the service itself because Tremor is designed to not create any load on your service as you are executing it. You have the option to run it as a standalone server where you give it the resources that it needs by itself. If that turns out to be too much of a burden for the application and you need to do a side car kind of model where you actually deploy it alongside your application directly in the same pod or sometimes even in the same container as a different process. It's lightweight enough that you can do that, which is part of the reason why we like it and why we built it is that we found that many of the other solutions for processing these signals just simply took up too much compute to be competitive with what we were able to create ourselves. And how long have you been with Wayfair? I've been with Wayfair for almost three years, three years in October, or it is October, so three years, hooray. Hey, happy anniversary. Yes, thank you. Obviously you couldn't have predicted the massive shift in all of our lives and maybe you didn't completely fully appreciate just how well-prepared Wayfair was for this crazy shift in all of our lives when you entered in. I'm not gonna let you pretend that you knew and that your IQ is 20 points higher than it actually is because you decided to go to a place that was actually prepared, but what is, can you share with us what that has been like? I mean, what is it like to be somewhere where the sky appeared to be falling and then all of a sudden the demands went through the roof? What was that like? That was a extremely chaotic but ultimately successful time for the company, obviously. It's shown definitely in what you can find in our stock and then how we've been doing the technology that we did very well during the pandemic. We were able to use the technology that we already had and be effective in adopting some more of the CNCF portfolio and some more of like cloud-native practices to make our infrastructure run better than it ever did at a time where we were in a crunch to be able to do better than we ever had as a business. And I believe that those two things are related. This is my personal philosophy for sure that I believe that the adoption of these cloud-native practices especially being pushed from the platform team that I work on were instrumental in being able to create an environment where developers can deliver value more reliably where then the experience of shopping at Wayfair becomes dramatically better and can handle the scale that you see when everybody decides to start shopping, everybody decides to start furnishing their home office. I was gonna make a joke that I bet that at least one of those things behind you was from Wayfair. You could be right, although some of the things over my shoulder are very strange movie prop type items. So, but you would be guessing correctly. I was very, very excited to talk to you from a technology standpoint because again, you guys were prepared for this. You can't respond in a way quickly enough that doesn't crush you unless you've prepared and you've got a framework that draws upon cloud-native technologies just as you outlined. So, against that backdrop, what are you seeing in from a technology standpoint in retail in general? Are you plugged into that much? Think of legacy furniture outlets trying to pivot into the world of cloud-native. Take your Wayfair hat off, your partisan hat in terms of competition off for a moment and talk to me about, if you're advising a fill-in-the-blank legacy retail store that's just dying right now that wanted to have an online presence from a technical standpoint, how would you advise them? What would you steer them in the direction of? I would definitely say that just-in-time engineering has actually served Wayfair extremely well where we're not over-engineering the solutions and using the big, fancy tools until we know that we need them. So I think that when we see businesses or we see anybody, any organization that decides to adopt everything first and then see whether or not it scales up, they don't see the results that they wanted because they're not using something that's appropriate for the size of the problem that they're trying to solve. So for example, if an experience that I can share from what we went through, I was part of and I've spoken and had some posts about like being able to scale up just the automation infrastructure for Wayfair where we were using a solution that worked pretty well but we didn't think about what was going to happen when it grew and we didn't react when it actually did grow. And so by instead reassessing, okay, we're a different size company now, we have different size needs for automation and different flexibility requirements to be able to use it effectively, we need to be able to adopt containers, we need to be able to support deployment into Kubernetes. How can we get there? And then continuing to evaluate that even during the process of building and during the process of making that available to the rest of the company. I think that if you're starting that process fresh or even if you're in the middle of that process, it's important to right size the solution and consider whether or not it's, if your online business isn't mid-tier but it's enterprise, then you need to build a system for that. If you have an online system that's actually doing not much of your business, don't build for the enterprise yet. Build for the size and then continue to scale it up as you go. Another thing that I just have to plug about the CNCF solutions is that they're incredible at being flexible to that scale that if you pick things that go from low to mid-tier then you can hand off from mid to high tier and then from high tier into enterprise scale. And I think that those things are available in the CNCF landscape and it's part of why we're excited to be part of it ourselves. So if you had a magic wand and you could solve one headache that you experience in your daily life from a technology perspective, can you think of anything that you would conjure up? Absolutely, I struggle and have struggled pretty much my entire career with being able to get like a good picture of the adoption of any given tool within the company. And I know that perhaps not every engineer that works with technology within a big enterprise firm has to think about whether or not their tool is being used a lot. Folks may not think about is a person next to me using Python or is the next person next to them using Java? Being able to have that kind of insight into what people use, how frequently they're deploying and how much they use it would be an incredible gain for us to be able to make better decisions about the platform of the company. Yeah, yeah. And you know, potentially the irony there is at the sort of tip of the spear of your business, understanding the customer, everything about them you possibly can is so important to give them a really, really good experience. And sometimes enterprises that have that know all sorts of information about me don't know what their developers are doing in a holistic fashion with a few clicks. I can tell you exactly how many orders I've ordered from a given online retailer in the last nine years, but I doubt they could tell me a lot about some of their infrastructure. So that's interesting. Well, what else can you share with us about Wayfarer? Anything that's not super proprietary and secret that you could share that's an interesting data point we were joking beforehand about, what is it, eight billion in your sales? Is it red staplers or any kind of factoids that people would be surprised about? Yeah, I think folks, something that's definitely not proprietary because it's literally on github.com. We just recently started putting a lot more elbow grease into our open source repositories. It's becoming Hacktoberfest now and we're very excited to be able to have these kind of more polished products, things that we expect people to be able to contribute to, where not even a year ago, it wasn't uncommon to have teams within the company that would open source their project and then they would completely lose track of it. And we had like some, we had to create new organizations to actually maintain them over time. So I feel like a really exciting thing that we're doing now is contributing to open source and earnest and actually getting developers' time scheduled to be able to dedicate to that effort. I think a lot of the biggest companies that are the most successful make time for their developers to be able to contribute back as well as being able to contribute just to the proprietary code that every company has to maintain. Absolutely, as human beings, we don't want to be toiling in obscurity and at what becomes a soul-killing exercise when you can actually get out there and have that sense of community, which is central to open source. That's great. It's a testament to Wayfair that they know that it's in their best interest as an organization to nurture that kind of talent within. So on that very positive note, I'd like to thank you so much for your time, Gary. And I appreciate the plug for shoes over your shoulder. And just want to say again, thanks a lot. Best of luck with your talk for KubeCon. And with that, I will sign off. Thanks for joining this Kube Conversation with Gary White of Wayfair in our continuing coverage of KubeCon, CloudNativeCon North America 2021. I'm Dave Nicholson. Thanks for tuning in.