Episode 32: The Hybrid Cloud Forecast - Outlook: Modern Computing

Speaker 1: You are listening to the Hybrid Cloud podcast where the forecast here is always compelling as we discuss real life challenges, successes, and stories from the journey to Hybrid Cloud with your host Andre Tost.

Andre Tost: All right, hello. Welcome everyone. Thanks for listening to this episode of the Hybrid Cloud podcast, and guess what? We have our second anniversary coming up. We've done 30 some episodes and over the course of two years and what I decided to do for this particular one is make it a special anniversary edition and we're celebrating by bringing our very first guest back on this podcast. Hello to Kyle Brown. Kyle was the very first guest I had on this show. It's still to this day from what I know the most downloaded of all the episodes, so it must've been really good I guess. Hello Kyle. Thanks for coming.

Kyle Brown: Thanks Andre. I really appreciate you bringing me on another time. It means that apparently we didn't screw up too badly that first time around.

Andre Tost: I completely agree. For completeness reasons, Kyle is an IBM fellow, he's also a vice president and his latest job is that he's the CTO to the IBM CIO and obviously, we'll talk about that some more. First of all, I recommend everyone listening to this episode to listen to the first one with Kyle as well, given that I don't think we want to repeat everything, but so we want to focus mostly on what happened during the last two years, and I think lots of things have happened, but maybe we can start out with a brief introduction and then more focus on what happened over the last two years, so to speak, and what's this job of yours that you have right now?

Kyle Brown: Let's start with picking up from where we left off last time. Last time we talked, the role that I had was I was leading application modernization for IBM and was a part of the garage, which has now since become IBM Client Engineering at the time. I'm still doing an awful lot of application modernization work. That's a big part of what my job is, but yeah, what happened is that in July of 2021, not too long after we had completed the recording, I ended up taking another job within IBM. I got one of these really interesting job offers that come along just a few times in a career and that I was asked to become the CTO of IBM IT, which is weird because I was flipping my role, I was now taking on that job that I had been advising our clients with. All of a sudden I found myself responsible for modernizing this portfolio of well over 2, 000 applications. At the time, we had an IT staff of about, oh, 10 to 12, 000 people that were working on those applications. We were also, in the process of a major consolidation effort in trying to bring down the number of data centers that we were operating on. I can tell more about that story later on but essentially, what has also happened during that time from an IBM perspective that complicated that entire effort, was IBM separated from Kyndryl during that time. We essentially took one of the biggest parts of IBM's business and spun that off into its own new company. That was something I found myself in the absolute middle of from the very first day of my job, having to deal with the fact that we're now disentangling ourselves.

Andre Tost: I guess I should know, but I don't, I mean, so is Kyndryl running their own? They have their own CIO, their own data centers or are we still sharing?

Kyle Brown: We are almost completely independent now. By the end of the year, we should be completely disentangled. There's still a little bit of remnants of things left in that we still have some folks from Kyndryl that are working with us at IBM to, essentially, help us with some of the IT that we do. Likewise, there's still some things like network connections that haven't been entirely separated out and little things like that, but we have almost completely completed that separation of the two companies and it's been a learning exercise on both sides. It was really interesting because I was just taking this opportunity to talk with our old friend inaudible who was asking about this and how both companies ended up using this separation opportunity as a modernization opportunity. We were both able to prepare for the future and get ready for new and exciting things by taking advantage of this shift and change that we had in the way our IT operations were being done, and be able to improve the IT responsiveness and IT operations of both companies as a result.

Andre Tost: I have lots of questions related to that in general, but I mean, to go back to kind of the initial questions and I remember that last time, now don't do it in every episode, but in most episodes I ask the guests about, give us your definition of Hybrid Cloud to start with and that kicks off the conversation. Obviously, we did that last time, don't want to repeat it all, but maybe we could go into, based on those experiences that you just described, has your view of what the Hybrid Cloud is evolved or changed in any form?

Kyle Brown: The answer is yes, it has. Being on the other side for the last two years or year and a half, and kind of witnessing the problems that IT organizations have from the inside, it has changed just a little bit, but it's a change in emphasis more in definition. Let me tell you the thing that I'm emphasizing now. Probably one of the hottest new terms in IT today is this idea of platform engineering. Well, what we have really discovered through the separation process of Kyndryl, but also through the modernization of our own applications that we've been taking on as part of this is that building a reliable, efficient, manageable, and especially highly automated platform that works in a hybrid multi- cloud environment is absolutely the key to succeeding in IT today. It's funny because this is something that going back to even my old job, I always knew there was this distinction between platform and application. That's like one of the basics of what we do when we are thinking about cloud adoption and moving applications to the cloud. One of the things that's become just starkly clear is how rarely people think about the details of their platform strategically. That really comes down to things like, well not only are you thinking about how your applications are going to run on the platform, are they going to be running on OpenShift? Are they going to be running on VMs? That kind of thing. Also, things like, well how are you commonly going to be managing your data pipelines across all of this? How do your applications that are generating all this data, how do they fit together let's say through a data fabric? Which is this kind of architectural idea of being able to track all the different pieces of data. How do they fit together with all the other applications and then how does that fit together with all the different data lakes you may have in your application space? Then finally, the other piece that has been really important in this is how does all this tie together through automation? When we think about automation, we usually think about application automation. I'm trying to automate my business processes and trying to improve that and make the whole process work faster or more efficiently or with less errors, but all of those same things that I just talked about apply to IT as well.

Andre Tost: I might even add one and say security because I have also changed jobs since last time we spoke and I'm now in IBM security, so that's all about data privacy, protection against attacks and things like that.

Kyle Brown: Isn't that important that you have consistent ways of doing that? Let me give you one example that's kind of interesting that we have learned as we've been going through our journey in the CIO. When we first came on board, and the interesting thing about the timing of when I came on board is that our CIO, Catherine Garini, had only been in the job about three months when she hired me. I was actually her first executive hire into her new department. What we discovered, one of the things that kind of just hit both of us really hard in the first few months is that we had had a number of application audits fail. This was a surprising thing. It's like you're coming into this new environment and you're getting the results of all these application audits and you're starting to look at them and going, " Well, why is this one failing and why is this one failing and why is this one failing?" We started to notice that they were all failing for the same reasons. In the end what it came down to was we had established consistency in the rules around what kinds of things needed to be done in terms of scans, let's say management of private data, in terms of all the different rules that we had set up. In IBM, we call this the ITSS rules, the IT security standards that follow. What we hadn't done is we hadn't really put the same amount of effort into automating those processes as we had into establishing those processes. The result being is what came back out of almost every single one of these audits was, " Oh, well this team forgot this piece of the standards or this team forgot this piece of the standards or this piece over here was done differently than best practice." They all turned out to be things that had we automated the process and had we made this part of the development process from the beginning in a standardized way, we wouldn't have encountered because teams would've been much more consistent. That's what we've been leaning into very hard for the last couple of years. We have been spending a lot of time and effort in coming to a common automation platform for, we call it the common CICD platform, but it goes way farther than continuous integration, continuous delivery. It's for continuous security scan and continuous testing and continuous checking and all the rest of the things that CICD want you to achieve in a common way, but a way that allows it to be parameter- driven, just in the same way that OpenShift is so that you start from the basics and you make sure that all of your boxes are checked, but then if you have unique requirements for your particular application, you can extend it without giving up on those same basics. It has just been a day and night change to what we've seen for those applications that have adopted this and so far, we're up to about somewhere over 60% last time I checked of our OpenShift applications have now adopted this common CICD platform. That allows us to not only be assured that all these checks are being done, but now we can correlate all that data. We can actually go into that and see the results of all these security scans and all the test results and all the rest of the pieces across all of these programs longitudinally. That is just creating a great opportunity for us to learn more about our process of the way we do things by just looking across all the different teams and then comparing the metrics that we're seeing there to the results that we're actually seeing come out of the data that's flowing into the data lake. It's cool. It's been fun.

Andre Tost: All of this seems to suggest coming back to where we started with definition of Hybrid Cloud, you said a Hybrid Cloud platform is key, but this platform is also, what should I call it? A multidimensional thing of sorts because it's a platform, you can have platform and infrastructure on middleware and even I would say there could be application platform services platforms. Now, you're talking about a DevOps type platform, so it's not just one coherent thing.

Kyle Brown: It is totally multidimensional and we've had to do that as we went through the adoption of our platform. Let me talk about some of the basics and then I'll kind of build up from the basics. The first thing we had to do is we had to have a really good sense of what infrastructure our platform is built on top of. For us, that was fairly easy. The easiest decision was, of course, OpenShift. We're going to be building everything on OpenShift because that is i IBM M strategic platform and I would say most of our applications, well over the halfway mark, running as OpenShift applications. Of course, you get a mix of monoliths and microservices based applications and there, but that's okay. OpenShift handles both pretty well. On top of that, or in addition to that, you have those applications that let's say for one reason or another can't run in OpenShift, you're going to need virtualization. There we've been able to leverage OpenShift virtualization as one of our solutions there. Cool thing is that ties straight into the automation frameworks of OpenShift and works out really well with that. In addition to that, there's other platforms too. Z being probably the most important of them. We are never going to get off of Z. I'm just going to make that bold statement. Z is going to be around forever because it is highly optimized for certain kinds of things and there is nothing on earth that can be as effective at that kind high speed reliable transaction processing.

Andre Tost: I think it's not just IBM, right? I think we can point to many, many large enterprises who've come to the same conclusion, right?

Kyle Brown: I talk to banking customers a lot still and they're all in the same place. They're basically saying, " Yeah, Z's going to be around forever," which is wonderful for IBM, but it means that you have to think about that as part of a platform that includes things like OpenShift. Well, there's one more piece. There's cloud.

Andre Tost: I was going to say how much is there a drive to move everything into the cloud? Because that's kind of what people do, isn't it?

Kyle Brown: It is and it isn't. What we've found is that cloud is the third leg of this kind of three- legged stool, if you want to think about it that way. There's kind of the OpenShift leg of the stool, which primarily for us runs on- prem, but also in the cloud. There's the Z leg of the stool and then there's the cloud leg. Well, in the cloud we're also running OpenShift, but what we're also taking advantage of in the cloud to a great extent are cloud services. On IBM cloud, we're really big consumers of DB 2 on the cloud. I could paint all the rest of the databases on the cloud, cloud object storage, but also we do consume things like the Cloud Pak for Business automation on the cloud and some of the other SaaS services we consume directly on the cloud. In addition to, of course, our hosting of our identity environment is entirely on the cloud using IBM services. The cool thing about that is it allows us to focus on the parts of the workload that we know run best in one place or another. If it's just an OpenShift workload, we can actually figure out where it's cheaper, easier, and simpler to run those workloads, be that in Rocks, which is running on IBM cloud or on OpenShift on- prem that we're managing or soon even on OpenShift on Z. Part of that was not only having that strategy of having these three legs, but then being able to gather the data so that we could do finops across all three of them.

Andre Tost: By the way, when you say OpenShift on Z, we can run OpenShift on Z Linux today and have been doing that for a while, but now you're talking Z OS, right?

Kyle Brown: Could be either one, actually. I want to emphasize the fact that the cool thing about OpenShift is it's OpenShift regardless if it's the Z OS or if it's the Linux. And part of us being able to take advantage of that fully like that is this idea of multi- architecture images, which is something that we're just putting in place now, which is a really cool feature in the latest version of OpenShift that allows you to build these things simultaneously and then deploy them as needed. That's just the basic level. Think about that as just having these conscious choices of what your platform base looks like, but then on top of that, you have to automate all of your IT processes that it takes to do something like adding a new node into an OpenShift cluster, or adding a new network connection, one part of your network to another part of your network or opening up firewalls in your application. Well, depending upon where you're falling in the three parts of that triad, that might be a little bit different. If I'm deploying a workload over here on Z, then let's say firewall management might be just a little bit different than if I'm doing it in IBM cloud, which might be just a little bit different than if I'm doing it on one of my data centers and being able to automate all of those and abstract that away so that you can just have common requests that know how to work their way through the automation is, I think, kind of that next part of platform engineering. It's being able to not just have a platform that's given to you by whatever vendor and whatever capabilities they have, but then putting this common layer of APIs and automations on top of it to be able to manage it in a common way and then being able to have the observability and the finops and the other management aspects to it to be able to make those decisions about workload placement, workload utilization, and the whole decision around when you run things where and why.

Andre Tost: That's an important point, I think, and I would assume that we touched on this maybe two years ago, which I can't quite remember, but where sometimes we give this illusion of saying when we have Hybrid Cloud and we say OpenShift everywhere, then we say it doesn't matter where it runs, but it does matter where it runs because especially when it comes to data, it does matter where the data is and how I get it there and where I need to ship it, because that's where it becomes expensive. I need to have that idea of, that's why what triggered my thought was when you said workload placement, but also data placement, data location is really important.

Kyle Brown: The cool thing about this is this gives you the ability to make choices about both your data and your workload separately when you've got a Hybrid Cloud platform model set up like this. One application that I'm thinking of that we just recently moved around, we are using DB2 on Z as our operational data store because DB2 on Z has got some amazing features in terms of its availability, its transactional performance, things like that. At the same time, the workload itself is really just a Java workload. That can run anywhere and running it on Z is not necessarily the thing we have to do, especially since one of the things we want to do with it and scale up and scale down a lot. Well, in that particular case it made more sense for us to do that one in the cloud. We have this workload in the cloud on OpenShift that's scaling up and scaling down all through the month and it's got this DB2 on Z database underneath it that's keeping all that data constant and is absolutely reliable and never falls over ever. That's just one example. I could look at similar examples to where even in one case we have some workloads running on Z that are actually pointing over at some data that's in DB2 Pure Scale on the cloud. That's because Pure Scale gives you a different set of abilities in terms of what it can do transactionally and the kinds of things that it can manage. Not to mention all the other types of data. Data storage approaches we could do like cloud object storage and all the rest of the things like that.

Andre Tost: Then there's also, there's money. I kind of picture it and you can correct me if I have it wrong. As the CTO, you're interested in technology, in modern architecture and getting the best out of IT that you can. Then the CFO comes along and says, " No, you can't do that'cause it's too expensive," right?

Kyle Brown: Exactly, and that's the problem that I as the CTO, Catherine as the CIO have is that our boss's boss is Jim Cavanaugh, the CFO, and he's always doing that. He's always asking us to do more with less every single year, you can guarantee that's going to be true, but that's where that idea of finops becomes really important. One of the things that we've been focusing on a lot is the ability to track costs all the way down to the project level and to be able to compare operational costs and runtime costs across each of the different legs of our platform so that you can make these kinds of choices about, well, in this particular case can I make a trade- off between let's say reliability or performance on the one hand and cost on the other? Because every choice you make of choosing one non- functional requirement over another has an associated cost element. That's where this kind of finops analysis becomes really, really important in making those kinds of choices with your eyes wide open.

Andre Tost: Interesting. By the way, I don't want to forget to ask, when you said we've been consolidating data centers, I assume those are global, they're not all in the US or are they?

Kyle Brown: That's right. They're not all in the US. This is an effort that actually started before I came on board. The first number I remember seeing of how many data centers IBM had dates back to maybe nine months to a year before I came on board. At one point IBM had 76 data centers that we had applications running in.

Andre Tost: For our own internal use, is that what you said?

Kyle Brown: Yes, but, and this is why I'm going to say yes, but, you have to remember that that's because IBM had GTS at that point as part of our business, now Kyndyl. Operating data centers was part of Kyndyl's business and they had lots of other customers workloads running in those data centers too. In fact, yes, we had 76 data centers that we were paying at least some portion of as part of that. Well, by the time I came on board, Matt inaudible, our VP of infrastructure who came from Red Hat and is absolutely fantastic, so shout out to Matt, had already reduced that number down into the forties. He had already been leading a consolidation effort together with the folks from GTS to just kind of say, " Well, do we have to be running in so many places," and had brought it down as part of our OpenShift adoption process as well. Well, now we're almost to our end state. In our end state we're probably going to end up with somewhere around four data centers and that's a huge change. Now, in all of those cases, those are colos, so they're actually in the same physical locations as IBM, Cloud and that shows how tightly those connections are between the parts of those triads. Our mainframes will be physically co- located with the IBM cloud data centers, which are physically located with the same racks that are running our on- prem applications. That allows us to essentially reduce our latencies down to zero, more or less, but importantly allows us to also, add that same level of redundancy into things. They're not all in the US. We have two in the US and we have two in Europe. That's the kind of the end state that we're going to now. We're not quite there yet. We still have some things running in Asia and some things running other places, but that's kind of the end state where eventually seeing things get down to when we're inaudible.

Andre Tost: Even though I would almost disagree and say there's never going to be an end state because things never end. There's going to be one thing that is constant in our industry, that is change, right?

Kyle Brown: Yeah, exactly.

Andre Tost: Actually, gets me to another point that I wanted to make in terms of, all right, so what happens next? What's kind of the outlook? I'm looking for are we heavily looking into things like machine learning, for example, or other AI technologies and how does that impact things that are changing? Are we looking into technologies like say serverless, we talked a lot about infrastructure platforms, but pushing this notion of a platform even higher up the stack to where we can run certain workloads in a serverless way and maybe you have other examples. How do you balance the need for consolidation with the need for constantly staying ahead of the game so- to- speak?

Kyle Brown: Part of that gets to your platform discussion. One of the things that we have done as part of creating our platform is we've made sure that we're giving people the opportunity to access opportunity these advanced technologies as part of the platform. One of the things that we do, for instance, is part of our OpenShift platform is Cloud Pak for data. We have Watson Studio and Watson Machine Learning and Open Scale all available there for people to use. We actually have an AI guild across the CIO organization that is working to develop best practices. I wouldn't even go so far as saying standards, but at least common approaches to being able to develop AI applications and machine learning applications. Now-

Andre Tost: That's good to know, by the way, and I'm going to make a note of that because I happen to lead an AI team within IBM security, and so we should at the very least compare notes.

Kyle Brown: Absolutely. I've got some old friends that you probably know from both of our lab services days that are now part of that. That's one example. Another example, you mentioned functions. OpenShift functions is also something that we're making available for people to being able to start use so that they can use serverless either in IBM cloud or in our on- prem environment, and we can just move it around and depend, figure out where that makes the most sense. I tell you the one that's got me the most interested right now, and this is an on the horizon thing, okay? One of the things that we've been thinking about a lot is something that was actually just discussed as part of the announcement that the White House made a couple of weeks ago around the president's cybersecurity initiative. If you read down through that cybersecurity initiative and you kind of go through all the different points on that, one of the things that you may have noticed is that the White House is saying, " Yeah, we're going to have to start thinking hard and making rules about post quantum encryption." Okay, what do I mean by post quantum encryption? Well, here's the deal. One of the things that we know how to do if you had a big enough quantum computer is you could run Shor's algorithm to break RSA encryption. This is well- known. I mean, people have been aware of the existence of this algorithm for years. The only problem is we didn't have a quantum computer big enough with enough qubits and enough reliability in the qubits to be able to actually run it on enough bits to be able to break an encryption key like that. At the rate that things are improving both in the reliability of the qubits, which gets down to quantum error correction, and in the total number of qubits that are available in our computers, we're guessing we're probably no more than five to seven years out from Shor's algorithm actually being able to break standard RSA 256 keys easy. To be honest, it won't be long after that just given the standard acceleration of computer hardware before, I would say, even bigger keys are going to be also just falling like dominoes. If we are predicting that on commercial hardware now in five to seven years, if you're a state level actor, my guess is you may be ahead of that curve. What people are worried about, and this was actually discussed is part of this announcement that the White House did not too long ago, is that people are worried about a new kind of attack that hasn't done in existence until now, to where people may just be recording traffic on the internet, encrypted traffic with the assumption that sometime in the future they'll be able to break the RSA encryption on that and to be able to read that traffic. It may be two, three years or longer before they can do it, but the assumption is they're doing that. That means we have to move to algorithms that are not going to be vulnerable to decryption using Shor's algorithm. That means we have to move to things like the Crystals Kyber algorithm, which is one of the candidates that the National Institute of Standards is looking at for a post quantum encryption algorithm. That will have to be incorporated into things like TLS. That's a big change. I've got what, 2, 000 some odd applications in my portfolio. That means I'm going to have to think about all 2, 000 of them and think about, " Well, what does this mean to each of these applications?" Well, in some cases it just may mean that the platform has to change, the new version of OpenShift that would incorporate one of these algorithms might do it. In other cases, I might have individual products that have to change like, " Oh, I'm going to need a version of MQ series that would support this." In the worst case, it may mean I have to actually have to go in there and change the code. I can't just change the dependencies to make this work. It's not enough to just pick up a new version of web CRMQ or anything like that. That effort could be as big as the Y2K problem, and that's huge. That's why we're already starting to think about looking at our portfolio and figuring out how to make these kinds of evaluations and then be ready for this when the National Institutes of Standard finally picks the algorithm that it's going to pick and vets it so that when those products come along and when we're ready to do that, that we're prepared for this kind of new future of post quantum cryptography. That's the part that's actually got me really interested now because near the start of my career, I lived through the whole Y2K thing and now something like 25 years later, after all this time at IBM, I'm to the point of having to go through yet another effort of this scale. That's kind of the thing that's got me most interested right now in understanding what that planning's going to look like.

Andre Tost: To allow a shameless plug, we and IBM security are very well aware of this problem and we talk about it all the time. In fact, we do have offerings in place to help customers like yourself in their decision- making process. Because just like you described, I think everyone's going through this kind of epiphany now and saying, " All right, now what do we do?" Part of the challenge is that just like you said, these candidate specifications are not final yet, but there are still things you can do. I don't want to get into, I'll save that for another episode maybe. Now we're rapidly running out time here. I have one more question that is completely unrelated to all of this, but I wanted to, didn't want to let you go before asking it is you said there was not only changes in your job life, there was also drastic changes in your personal life, so is that something you want to tell us more about?

Kyle Brown: Yeah, it's interesting. Here's the deal. I may have even have talked about this the first time, but one of my hobbies is running. This is something that actually started for me way back in 2014 to where I had a health scare around diabetes. One of the things I had to end up doing was changing my diet and changing my exercise routine. I took up running at the time and it turned out I loved it. I've now run three marathons, probably a dozen or more half marathons, probably closer to two dozen and more 10 milers, 10Ks and 5Ks than I can count. In a week, I'll run-

Andre Tost: I can share, we're connected on Facebook and I always see these pictures with Kyle under the Eiffel Tower or Kyle in Dubai in the early morning hours before meetings start and getting in a 10K before work.

Kyle Brown: Exactly. Always wearing running gear. Something really interesting happened, interesting in one for instance, this July is when I first noticed it. I was getting ready to train for the 30th Anniversary Walt Disney World Marathon. I love Walt Disney World marathons. It's just a great place to run a marathon. I had just started my training and I started noticing that I would get about two miles into my routine and I would start to feel a little bit of discomfort. It was weird because it felt maybe like my asthma was acting up. I went to my doctor and we tweaked my asthma medication and helped for a little while, and then I went back to training and came back. We realized it wasn't my asthma, and that led to a whole bunch of additional tests we ran because the first thing that we thought is, " Well, okay, it's probably your lungs might be your heart, probably isn't your heart. Tell you what, we'll put you on the treadmill to be sure." They put me on the treadmill and my heart was perfect. I had runners spikes on that the way you want it to look. It took a long time for it to get up to the high heart rate and everything was fine. That was good, but we kept thinking something might be going on. I went to see a cardiologist and he does the same thing. He runs the chart, my heart looks fine, puts me on the treadmill again, runs the chart, that looks fine. Then he goes into, " Well, okay, there's a couple more tests we can run here." He does one to where he runs a scan of my heart where it's basically using a tracer and kind of taking a picture of it, CAT scan like that, and it looked fine. Then he put me on the treadmill, treadmill was fine, ran the second scan and it's like, " Oh, hold on, there's this little spot on one of your arteries." Then they said, " Well, okay, that could be the beginnings of a blockage, so what we're going to do is we're going to run the next scan," which is actually the first intrusive, their invasive scan, and that's called a heart catheterization. That's where they actually stick a little thing in one of your veins and they thread a tool down into your heart and they inject dye right at your heart. Then they can watch all the blood vessels, go and kind of watch the blood flow and they can look for a blockage that way. If there is a blockage, they can even right then fix it by inserting a stent. Okay, so that was the plan I thought my cardiologist saying, " Yeah, yeah, we'll be fine if we do find something, we'll insert a stent, we'll send you home the next day. You're all good." I go in for this and you're awake through the entire procedure and there-

Andre Tost: Is it painful? It doesn't sound like fun at all.

Kyle Brown: No, it's not painful at all. You don't even notice what's going on. It goes in either through your arm or through your leg. They numb that spot. It's no more painful than just getting a cavity filled in that respect. You can't really see anything. You can hear them talking in the background to each other as they're threading things around and doing stuff and they just keep working. Eventually they say, " Okay, we're done." They wheel me back to the room and then the cardiologist comes in and he talks to my wife and I and says, " Okay, we didn't insert a stent. Turns out you didn't have one blockage, you had six, and that would be inserting too much metal into one heart." He said, " The next option is you need a sextuple bypass operation."

Andre Tost: I don't think I've ever heard of such a thing. That must be rare.

Kyle Brown: Neither had I. After we got over the shock, we consented to it and about two weeks later I was wheeled into an operating room and they bypassed six different locations in my arteries feeding my heart. Now, it turns out this was all genetic because the first thing people ask is, " Wait a minute, you had a healthy diet and you're a marathon runner. How could you have six blockages?" One of which was at 90% already. The answer is that it's genetic. Even if you do all the right things, it turns out that with some genetic predispositions, it's a 50/50 chance of whether or not you're going to develop a blockage. It turns out I was on the wrong side of that coin flip. They took a vein out of my leg, turns out that you have redundant veins in your legs because this is the thing I was concerned with. It's like, " Doc, will I be able to run marathons after this?" He's like, "Yeah, yeah, yeah, yeah, you're fine. Don't worry about it." They took a vein out of my leg, they bypassed six different arteries in my heart and then I was out of the hospital, I think, three days later and I had to-

Andre Tost: But the problem got fixed?

Kyle Brown: The problem is fixed, but I had to restart everything. I ended up taking a couple months off work. The good news is now I'm not only back to work, but I'm back to running. I actually ran 10 miles this weekend.

Andre Tost: I was going to say, because I didn't see pictures fairly recently, so I know for a fact that you're running again.

Kyle Brown: In Dubai. Just a couple weeks ago. I ran a 5K with a bunch of folks from IBM and Red Hat. The good news is I'm back. I flew through physical therapy because I was very highly motivated to get back to running, and so set all sorts of records on half fast I managed to get through that, but I'm back. I am, again, watching my cholesterol and my diet even more carefully than I was. It's been an interesting experience through this and it really makes you think about the things that are really important and makes you think about the legacy you leave behind in a case like that.

Andre Tost: Wow, and so for a while I was thinking why did I have to ask? But now I'm happy I asked because obviously, there's a happy ending to all of this.

Kyle Brown: Absolutely.

Andre Tost: Talking about endings, we've come to the end of this episode. Thanks again for coming. Maybe two years from now we'll have you back and see what happens in the next two years. Right. That would be appropriate I guess, but for now, thanks a lot for joining us today.

Kyle Brown: All right, thanks Andre.

Andre Tost: All right, so with that, we'll wrap it up for today. Thank you all also for listening and hope to see you all soon. Bye.