How Chevron is developing an agile dev culture as they migrate
Welcome. You. Guys having a great build. Yeah. Yeah okay let's have some energy here hey so we, got Chevron, with us today Chevron, is coming up on their hundred, and fortieth. Anniversary, one, hundred and forty years so, here's a deal when you got a company that's been around 140. Years that's, a company that can navigate some. Twists, and turns okay, and once, again Chevron's. Navigating. A big shift, and that shift is digital, transformation. And so we're, lucky enough to have Victoria, Harris with us here, Victoria, is what, Chevron calls the cloud lady, and her. Responsibilities. Get this get this her, responsibility. Is to, take six, thousand. That's, right six thousand. Applications. And move. Them to the cloud. So. She's gonna tell you how she's gonna do that and joining. Her I've got Ryan Ram, Sharan he's, from Microsoft, with our Azure. Adoption. Team and. All. Right thanks Jeffrey so. My name is Ryan Ram Sharan I'm with, the team we call Azure, customer, advisory or better known as Azure cat and so. Who. Am I and what do I do so, about, six months ago you might have read in the news at Microsoft, formed a partnership with, Chevron. And so what does that partnership, really mean well it. Means a, commitment. To, the platform, and by platform, primarily. Azure. Across. A number, of different facets, so that's everything, from, OMS. And the operations, management suite to, analytics. To enterprise integration. And so. For that commitment, and this, partnership that we're building our engineering. Team said hey look you know we want to be in. The in the trenches, as they. Migrate, as they adopt a lot of these technologies, so we're gonna Park you. Meaning, me out there and make, sure that as Chevron. Adopts these technologies, we. Are ensuring, that our the, feedback is driven into our product teams to ensure that we're meeting or exceeding you. Know the solutions, that. We're being asked to provide and so. For. This Azure force commitment, again there's, myself, and a number of architects that are spend, our day living. Out at Chevron working. With them across this digital, transformation journey.
And So. Geoffrey, mentioned there's about 6,000 applications, that were moving but, that's just, on the migration side, it's, also a number of Greenfield applications, across again at things like analytics, so. You know leveraging, products. That are coming out to do interactive, queries and, all of that stuff and as well we've, got also things like enterprise integration right. So service plus logic, apps, building. Enterprise, integration for, all of the transactions, that run through all of the ERP systems, etc so it's AB it's a big commitment and again, what, we're there to do is help them as. They architect. These things and then again provide the feedback into our engineering teams so, with, that I'll turn it over to Vicki thank. You right they, think of that one back to Jeffrey, so. For all of you here this was a 45 minute session so you're in a 75 minute block so, we, don't have 60 minutes of slides so that's lucky for you but, you do have my commitment I will stay here the. Whole time to answer any and all questions that, you have full 75 minutes so we. Do want to keep the presentation short, this, is a very broad topic and the questions can come from all sorts of places and so I'm going to touch on things. From a billing, to DevOps and things like that so I want to leave a lot of time for questions. With. That I know I have a funny title public, cloud manager, Chevron, calls everyone a manager, I, did, start this journey about, two-and-a-half, years ago I think with one employee I have, completely, lost track, of how many I have now I think we just on our agile teams alone we've got one, 150. So. We, grew pretty quickly, which. Is a difficult thing to do in an organization, of our size but. It's. Testament. To the speed at which we wanted to move and some of the early successes we've, had and being able to move at speed. So, just really going to talk very briefly about chevron, diggie you an overview which, gives you an idea of some of the challenges, we are facing. Jeffrey. We don't like to I don't like to remember that we're 140, years old but, we. Do have a lot of legacy technology. And, so there's a lot of good and bad history there in terms of what we have to overcome. And. Really, mostly what we learned because. We've been moving very quickly we, have a lot of good learnings some, things were easier and some things were harder. With. That just a little bit about Chevron. We. Are a global, energy. Company. Operating. In over 30 countries around the world we. Have over 50,000. Employees we, also have a lot of joint ventures and strategic, partnerships, so, in any given day there's, over 150,000. People working around the world in. The name of Chevron, and our, main goal is to make sure those people are safe every day. For. IT we have about 4,000. IT employees, and there globally, distributed. Distributed. As well and. So those folks are supporting, our business units. And. As Jeffrey mentioned we have 6,000, applications. Anything. From an HR application. To. Something that's looking for a crack in a pipeline to something. That's managing, the liquification, of natural gas to, a. Basic finance application. Basically. Whatever it is we've, got an app for that we. Got to move it. But. It also creates, some challenges in, the distributed nature of our workforce if you, can imagine all these local solutions, are great, for the business they add value.
But. If you're looking to move into the modern world where. Everything is automated and you have no base of standardization. And. You do not have very many common, processes, that is a tremendous, challenge. So. Where we started with some of these challenges, we. Do have a very strong safety, culture strong, reliability, culture, which. Means we can check once we can check something twice we're really good at checking stuff we might check it 15 times it. Means we're highly, highly reliable, it, also means we're not very nimble we're not agile, and. We do not deploy. Code to production, very quickly. So. Also. That's a mental challenge, or a culture, challenge, when you're getting trying to get people to do something differently and, they, say I have all these manual, check check check processes. Telling. Them no, machine can do that and it's all going to be fine and it will be highly reliable sometimes. They don't believe you and so proving, to them that. It can be more reliable, it's. Also a challenge, but. That's critical, that's a mental hurdle. To get over. Is. The, cloud secure is it, reliable, we certainly had to prove, both of those things to. Help our business partners understand. Also. Early on we had high demand, but. The demand is I don't really know what I want but, I know I want cloud and, I want it tomorrow. So. We, have a change management challenge. On our hands, which is that people are asking for capabilities, where they don't understand, how they work, and. The mental model that they have is their legacy technology, and so sometimes they don't even ask the right questions, just, say I need that and I need it tomorrow, so. Sometimes you have to just walk them back and really try to educate, them that's, a tremendous challenge as well. Fixed. Legacy, costs and technical debt. If. You're only looking at a project I. Mean. I saw it a hundred times probably. In the first few months of my job, it's. Cheaper. For. Me to just maintain this thing than it is for me to transform, it and so your transformation. Case or cloud migration case, from a business case perspective, can, only be done in aggregate, because. If everybody everyone, is one offing these decisions. They. Tend to bias back towards the legacy, let me just invest a little bit more on the legacy. Also. You need a strong technical foundation. Network. Was our identity, buyout was our boundary. We. Were not really networked for the cloud we, had some SAS applications. But. We really hadn't designed or architected, for what it's like to be running the majority. Of your workloads, outside. Your own data centers just a complete different shift so, we had to invest a lot in security, and working. Very. Little. Or capability, certainly. Very little technical capability. But. Getting broad. Misunderstanding. About what cloud is what the technology can do so just kind of low capability. All around in the tremendous. Education. Challenge. So. What do we do about all that well. We started with something that was really just kind of a how do we get started we called it the cloud center of excellence. We. Did anything from writing master, legal contracts, that people could use for SAS providers. To. Answering questions to, doing basic architecture. Governance. We. Also started to build out the foundation, for the security, and the networking, for. Multiple, cloud providers. At. That time, we. Didn't address some, of the major challenges we didn't address the billing we didn't address migrations. And. So we were I would, say still testing, the waters, we're, doing with a lot of people doing a lot of proof-of-concept. Does. It work let me play with this let me test this out and, so, that's all well and good until. You actually really. Need to do something at scale and then you realize, kicking. The tires window-shopping, whatever, analogy you want to use at. Some point, you have to stop pio seeing stuff and you just have to go. So. That moves us to what we call the cloud program. That. Was our response, to the leadership team saying hey, why. Aren't you going faster it's. Like okay. We can go faster here's what it takes to go faster. We. Need some funding, obviously, but we also need a different structure. So. We stood up a program to change our organization, so we literally launched, a new org design and we put people into that org design.
We. Defined, how our services, would be managed and operated and. We covered all the super, fun stuff like internal. Billing, which. Actually, is. In. The early days was fifty percent of, the questions that I got, so. Nowadays I think a successful presentation and, you, guys can ask me that but, internally. Is when nobody asked me about billing, anymore I'm pretty excited. But. We had to take care of all those kind, of meat and potatoes of operations. For, a global enterprise a, lot. Of compliance, a lot of questions, about hey. If a partner is hosting, this what does this mean to import export compliance, law things, like that so we had to crank through all of that work, and. I. Would say if. I could draw a bar there in, transition. To the last little cloud there we, started running agile. And. I would say that was the major. Accelerator. For. Allowing, us to get going did. We do great at it no. But. We started, and we got better we. Got better and, we got better the next time we're, still getting better and, we get better every day, so. Any, of you I would, say one, of the biggest, keys and it's in our critical success, festers is running, agile we use safe. The. Scaled agile framework. But. Just getting started, even, though you know you don't, know everything. Standing. Up the teams to, just start migrating applications, we, did that in advance of having our master plan because we like to have a master, plan for everything but it was never as right as we wanted it to be because there's so much uncertainty and. Costs, the. Changing nature of applications. If. You're, flipping something, to path services it doesn't really look like the same application, when you're finished how do you cost all that stuff so we had so many questions we. Couldn't answer but, while we were trying to answer all those we just thought let's, go let's. Start learning and building our patterns, and so, that was one of the best keys to success, so. Are we fully optimized, no, but we have product, based teams we're, moving quickly we're. Moving 150, apps this year. 500. X in terms of modernization and then we're lifting and shifting the rest and then we put them into a modernization, queue, after they get lifted out. So. A little bit about what we learned. It. Was 2012. I think we had our very first Azure, subscription. But. At that time I would say two people, were. Using Azure, and. We lived in that I would say until about 2016. I, think. There was a room I would say was about double the size and we had one of our senior executives, actually say over my dead body does. Our data go to the cloud so we had a lot of educational. Challenges, also. It. Wasn't. As well-known. There. Weren't that many stories, about how cloud was secure how. Safe it was how reliable, it was you. Didn't start to see those come into the marketplace in around 2015. Certainly. By 2016. Those stories, those anecdotes that, our, CEO or, our CIO was hearing or reading those. Became deafening, in, terms of what other people were doing that, helped kind of pressure us and change our approach. And. Certainly by late 2016, early, 2017, people. Were saying why. Aren't we going faster, so. Really quite a paradigm. Shift there, in terms of the way our organization. Felt. About it and asked about it, and now it's. My boss usually says are your feet on fire because.
We Have so many people saying I needed it yesterday can, I have more, faster-faster-faster. So. Where, I used to get the I don't. Need to talk to you don't. Ever touch my stuff now. We get please, can you help me and. We have so many of those requests, we can't accommodate them all. But. Some things actually turned out to be easier I think. Our going in assumption, look. Looking, at our infrastructures. Of service workloads and our password clothes, and, when I talk about our, situation our, six thousand applications roughly. Two-thirds, of those are vendor applications so. Those were clear I as workloads, all, of our custom developed, applications. We thought would be easy conversions. To, server. Lists, path services and, that is still our intent to convert as any of those as possible. And. So when it came to a complicated. Some. People have done very complicated, analysis, in terms of workload placement. And things like that for us it was pretty easy, most. Of our custom, applications, are dotnet and so most of them are going to path services. And. Why, was that easier. Well. It actually turned out infrastructure, automation for. Any of you who can't build McKenzie's presentation, it's really hard. We. Weren't prepared for it and we didn't have that capability for, it and it just took longer than we thought and so, to actually drop the infrastructure, completely, was easier than we thought not to say was we. Didn't have challenges, but, we were students certainly able to move more applications. That way quicker. Also. Like. What I just mentioned demand-pull I. Don't. Have to go around with, my hat begging, people to do stuff anymore, so. It's a flood of demand so, we have everyone raising their hand could I go next can I go next so certainly. You have to prioritize all, that and you have to plan that but, it's different story. Large. Enterprise, IT projects. Many times you're. Competing for priorities. You have to use, you. Know more heavy-handed metrics, you must do this we. Don't have as much of that because our. CEO. Is very interested, in digital transformation. Everybody. Understands, how important, this is and. They I want to go first basically, so our challenge is all those people want to go first is how, do we order them. Again. We weren't great at agile, when we started. But. Once you get started it's, not that hard you get some teams going get, some coaches if you need them again. One, of our most important. Success, factors, we. Wouldn't be where we are without that, also. Getting started on the pipeline, not. As hard as we thought I. Don't, have the architecture, diagram, again. If you pull it from one of the presentations, yesterday, our. Pipeline, diagram, once, we got that sorted out and understood what needed to be automated, and when, how it was gonna work. We. Could start moving stuff and, that was pretty cool.
An, Empowering, developer, is is one of the most beautiful, things I've seen, people. That might not have touched their code. I'll, just go on do something else like it's, so easy now I can make it better and like Oh someone so is doing something why are they doing that well because they could and, it didn't take them very long. And. So that's what. We wanted to do but to actually see, it happening, is very very exciting, because. Sometimes you do these things in concept, yes, we want to enable people and then so when you see it actually happening, where, they don't have to go through some central organization. And they don't have 15 tickets to, get a piece of infrastructure. They. Just deploy what they need it's. Really beautiful. What. Was harder than we thought I. Every. Day I kind. Of underestimate. The culture change. It. Doesn't matter who you are sometimes, that you deliver, the message 20, times 25. Times. Sometimes. It still doesn't stick it. Doesn't stick because people don't want to hear the change message, uh and. It's not always the infrastructure, folks. Sometimes. It's developers, sometimes. It's finance people. They. Just have a lot of there's, a vested, interest in. Not. Changing, I want to do it the same way so. I'm. Not going to do anything about it until somebody tells me I absolutely, absolutely, have, to and, so. The. Feedback I always get is Vicki your team needs to communicate more. Like. Okay I have this, budget we have this many slides we have all these communications. Meeting is they never ever enough, because, this is such a big change. Certainly. For our tea folks we're changing everything about the way they do things and. And, that's been hard and as, I mentioned infrastructure, as code has. Been a little bit more difficult, but part of that is the, same as below it's security. Those. Two challenges, were the same and the reason I say that they were the same. Is because, it's all manual. So. Imagine, if you have a request, and I just have tickets, I have 20, tickets that are created by this request, and they go to 20 different teams first of all nobody can track all those we don't really know where your request is uh.
But. Also if, you need to automate that you, may have to go to 20 different teams and negotiate with those teams where's your script where's your API need to pull that work back that. Is no small task that. Was really much much harder than we thought it would be and. For security our challenge, was I, would, say even greater, because. We'd made a significant, investment in security, because. Of the we'd. Really made a big investment in our cyber defense but. We made that investment, at the time where we were securing, really, our network perimeter and not thinking about an identity perimeter. Because. It wasn't as large, a priority, for the corporation, this cyber, defense, was, a greater priority than cloud we. Kind of knew that at the time but we also knew we'd have to unwind that and that's, been kind of a beast to unwind. And. Again automation, of all those manual processes, there were so many I don't think any of us knew how many there were and. Everybody. Who's operating, one of those, even. If it's I've bought a ticket. My. Job I run the script I'm gonna press a button to run the script and. Then I'm gonna log back in tell. You I did it. So. We. Had a lot of that I'm a little embarrassed to admit but. By the end of this we'd like to have none of that because. It's just not necessary, but, part of that is also the culture change of teaching, people your, value, is not there, your. Value, is in writing a script to, automate a new process, and then a new one and a new one and. Keeping our levels, of incidents, down your vow that's not your value is pressing, that button and telling, us that you did it. All. Right. So. In terms of what's really really, really important, Oh. Telling. People that they have to automate has been hard because. A lot of times it's just like I. Let. Me do it just this, one process, and, so in some cases. I've. Just costed, it out for them and said, this. Costs over, the life of this that, one manual, task is gonna cost fifteen million dollars, if you if you look at it properly and, so. I will. Keep your task manual, if you will give me fifteen million dollars. And. Strangely. They don't want to give me any money. The, next question is well I can't, do it could take six months it could take eight months so we would usually say will. Automate it for you it's, still yours and then we can give it back to you and. Then we would usually get the responsive, I, don't. I want you to doing my work I'll figure it out so, but again that. Teaching. People what's. Always have been acceptable, that what has always been fine it's, always been fine to say you need to wait eight months for this and so to come and tell someone sorry that's not fine anymore I'm not accepting, your answer. That's, a little bit of a shock for them but, it's critical. I mean there's no way we can scale with all that manual stuff no way I already. Talked about agile. And. I'm gonna say it one more time if. You're not great at it it doesn't matter just, get started, because, it adjusts itself. One. Of my favorite other companies, I visited, one. Of their teams said we do scrum Khan full ban. So. You could adjust the process we, use a scaled agile framework, but. We're not that religious, about it there are some things we do really well and there's some things we're less rigorous about so. You have to adjust it to your own processes. But, if you don't do it you can't move at the speed that's required, and. So we have had dependent, projects. That we're running what waterfall, I mean that's what we did at Chevron for years, we were very very good at it but. None of them could meet our deliverable, timelines, none of them so, I watched them fold, last. Two years one after the other one after the other and, every, one of them had, to start running agile because, they just couldn't, we. Can't afford first, of all we can't afford you to take, two, months to, come, up with a set of three recommendations and, then maybe whittle them down over another many.
Months. One. Of the reasons, is that, your answer is gonna be wrong anyway, because. Technology. Is moving so fast that, if. You take a traditional project. Your. Work product, is actually obsolete. By the time you deliver it that's. Not a nice message for people to hear either which is hey, how's. Your project your. Deliverables gonna be obsolete when, you get to it. So, the. Better way to do it is to have them come sit with some of the other tank teams, put. Them in your stand-ups, have them get going so, that they can see hey, this is a faster, way to deliver and I. Can, change my deliverables, to fit the pace of the technological, change, that we're, facing I. Already. Mentioned strong network and security but. I will say from. A priority, perspective, certainly. There were many reasons we chose to partner with Microsoft, is our primary cloud provider, so. I wouldn't say this was a major decision criteria. But, is, it I was, relieved. Because. Double, threading. All. The networking and, all of all the security so putting, out investments. For multiple cloud, providers, when. You're getting the same services, from them. Didn't. Prove to be very economic. For us so, we need our resources, focused on one, so. That's why we're focused on Azure, and. It's allowed us to move faster as well so we're not trying to put all these resources, to. Build. Out multiple, platforms, just in case we might need them. And. I would say adjustments, to our organization, were critical, because the culture change is so hard. You. Can't just have a few people working on this and not touch your organizational. Structure, that's. A probably, a way to fail, so. You've got a pull out the. People who are willing to run fast the. People who are software engineers the, people who can really deliver and you have to protect those people so. That's, part of my main job really, is to, protect, our agile, teams. Shield. Them so, I take all the grief all the politics, all the junk that comes my way so that they can keep working. So. If they're in an org structure. Where. The legacy, processes. Keep. Intruding, on their work you're. Not going to survive you're gonna have to pull them out I, don't. Want to say we got it right the first time we do understand, we'll need to continuously, change. But. We did. Consciously. Change, our organization. With the understanding. That we, could not leave some of these groups in the same group that we're supporting the legacy, technology, do we think will merge them in another couple years yeah we probably will but. We need to give ourselves some, room to grow and, really incubate, the capability. Build. Oh see okay so that sounds really easy, it's. Super hard. But. It's critical so, you can't give up it's like the change management you just have to keep working at it you have to hire it you, have to build it you. Have to do buddy systems. That's. One of the things we have working on our agile teams is we have a team buddy, that's assigned to the new team once we train them up and get them going so that we can help build capability. We. Have some of those people sitting over there they've. Been awesome oh. Thanks. For asking Jeffrey Oh see sorry, that's a chevron term organizational. Capability. Skill. And. Knowledge, and. So but, that was across the board so we have programs, in terms of training. But. Again all the formal, training is, not as helpful as the, hands-on, training for the people who are doing the migrations. And. Then there's broader, training. That we need for, our entire, organization, because the interest, level is so high and for.
People Who are maybe not ready to actually do. Sandbox. Environments. Help them learn just, to get them interested do. You want to do this do you want to not, here's. Some things you can play with we provide that type of training, and. We also provide, kind of more generic, training so. Part. Of our organization can, understand, where they might fit in the new world, for. Their role and. Then, lastly services, and commercialization. That, gets back to the can. We internally bill it but, it's also do we have the automated, tooling. To. Manage ordering, all the way through the provisioning, to, billing, because. That's that's, where we want to be we don't want to have manual, processes. In the middle of that and so if you don't address how all that works. It. Makes your life a lot a lot harder and like I said that's kind of the unfun part of cloud but it's got to be done. And. So this is my plug for, we're hiring developers. As. Many as we can we'd, love to see all of you work for Chevron but I know some of you need to work for Microsoft but, we're really excited about where we are we're excited about the capabilities, we're building. We're. Excited about the people who are coming to work for us we. Have a lot of great work ahead of us. And. So with that I'm, gonna hand it back to Jeffrey, all. Right so what now is Q&A by the way so the first cue is somebody. Wants you to take you up on that how do they get, in touch with you I think, my contact, info. Is in the build. Just. I'm on the site you, find me and you can find me on LinkedIn as well, excellent, so we have two microphones here I'll, start off with a few questions but then if you want to line up to the microphones, to ask some feel, free so. Victoria you know I've really enjoyed this conversation. Your. Journey with DevOps mirrors, quite a few other companies that I've heard the, one part that I didn't hear was that you know you start off with the Coalition, of the Willing. You have the resistors. And that big you, know a kind, of cultural, battle but often what I hear, from other companies, is the the, folks the first ones through the knothole the first ones that adopted, really. Kind of have a transformative. Experience, that. They they. Are delivering better business, results and, they feel that their career they're having more fun at their job have you found that absolutely. So. Certainly I've heard from more than one person that this is the best job I've ever had at Chevron of some, of the people working on those teams. But. Again those, are also the people who are receiving the kudos from their business unit look what we built so quickly look what I did. So. When. It has nothing, to do with me because they're all self enabled. That's. The greatest part of all and some of those people were also I, don't. Know if they were resistors, but I would call them skeptics, so, some, of the folks who are the biggest champions, now were early skeptics, of like oh come here I'll. Listen your training, I. Guess. And then literally, within weeks saying I got to get my best people in here we want to do this yeah that's one of the most exciting things is that when they when they finally make that that, leap. They, find out that boy they're delivering better results.
To Their customers, the business improves, and they're just having fun so. Let's, go, to our first question here. I. Can, really relate to a lot of things you talked about my. Company is only about 80 years old and a lot of our legacy code is no joke 50 or 60 years old written. In languages, you have never heard of how. Are you how. Are you. Dealing. With some of those really, old legacy applications. And. Modernizing. Them when, maybe. The development, team retired. 20, years ago. So. It depends on how old they are in. Some, cases they. Are literally not worth touching, if. Somebody wrote it 20 years ago and it's still running then, we would leave it be and not modernize, it otherwise. You probably have to rewrite it okay. But. Wait there's have common. A turn here if you're not familiar with it it's worth getting in focus it's called the strangler pattern. Right so in design patterns, it's the strangler, and what you do is you've got a legacy system, and, you want to get rid of it so what you do is instead up do not start. From scratch and rewrite, the whole thing that's death, what you do is you put something in front of it and then, have all the systems talk to that thing in front of it now the first implementation of, the thing in front just, calls the back end but, then step-by-step, you make that a thicker and thicker layer until, eventually you strangle, the backend that's that's the pattern that's the best practice yeah, that's pretty much what we've been doing thank you by, the way what was the what with the language give it give it up. One. For example Rocky, Mountain basic ok, you went on that one anyone. Which. One Rocky Mountain basic, anyone, anyone anybody, no no, yeah. Was a HP's. Implementation. Of basic on hp-ux, Wow. Wow. I, wrote. I wrote the first true, mumps compiler, so I was hoping you were gonna say. So. I had, a question about your your, program managers, and your, you know your PMO so. We're. Going down the same path you guys did and, it's. A little bit of a challenge they don't know quite what to make of all this they're used to being able to have their calendars, and their power you know and they're the other project, files and you're just going right along and, we're. Doing agile you, know for getting, things out in iterations. But. We can't give them a date like with that feature we don't have a date for that feature it's, done when we, get there right do. You have any and. You know anything, you could share about that. I mean I'm sure you went through something like that do you have some insights into that I'm, sure I have that conversation every, day of my life I, don't you know the answer to that. So. Part of it is you do have to have an overall plan. So. It doesn't man, running. Agile doesn't get you away from that you should know what your end goal is whether its financial, process. Oriented, number of applications, so we have that so we know where we're. Driving towards, and. So you have to put markers along, can I measure this can I measure that and. What we've had to train, our leadership, on and you know if you're using some, sort of classic waterfall methodology, is, because, we have a phase gating methodology, and so. Our assumptions, when we were doing waterfall, would be my estimate, on cost, and, schedule, it's gonna be really tight by this. Phase. And time and so, what we've done is saying it's. Just not gonna be that tight my, range of uncertainty, is greater and so here's a greater range of uncertainty that their work you're working in and so, I'm not gonna give you that but I'm going to give you this range and this is where we're driving and then they then.
They Back off a little bit and say okay. But. But literally opening, up that range of uncertainty and making that visible in terms of cost, and. Schedule and, deliverables so, that they can see okay, these people are not off the hook from an accountability, perspective. But. We also can't be too prescriptive about. What they deliver and when they do it does. It happen. I. Great. Talk by the way very inspiring my, question is around data so, one of the things that that I struggle, with figuring, out is, the. Data. Gravity. We. Have all our legacy, systems, on. Premises, in in IBM data centers, and. Well. We are building everything noon in cloud, however. We need access to that data and there's. Multiple layers in this challenge one, thing is just, the. Technical challenges, and then, we also have. Probably. More ownership. Challenges. So, letting go and sharing, your data with cloud applications, so. Have. You any insights, in this area to share I. Might. Ask Brian, to come up here and talk a little bit about analytics. But. Certainly. Integration. With, our data. And. Also. Some, very very old. Systems. Of record basically that are literally. Huge. But. We've also found as we move out some of the ancillary systems, that we're not as tightly coupled as we thought so, as you're building your. Migration. Plan some. Things that, connect back to an on-prem, data store you don't have to manage moving everything all at once there, is a way to pick it apart and, so we looked at some of those clusters in, terms of how we do that versus. Trying to over plan and like take everything and every integration, point and move it all at the same time but. You will have to do some D, architecting. As you go, and. We're also using the opportunity if we can use change, some of our classical. Classic. Point-to-point interfaces. To. Api's as we do that we want to look for those opportunities but. The, data is absolutely a tough challenge in some, cases like you said it is, easier. To. Build new in, the cloud, let. People migrate to that over time and then kind of let your legacy shrink, but. I wanted to get right do you want to say anything about what we're doing on the analytics, front in terms of data, front. So. I'm the analytic, side of the house you know we are moving. Want. Their big data, lake that they have on premises, into, the. Cloud so. I would. Say that there are certain applications that we can move the data but then others, are, a little bit more difficult, so like in the high-performance, computing world right, where you've got workflows, and those workflows have large very, large data sets seismic, data sets that need to move between that workflow well, an application, like that it's very difficult, to move just a piece of it to the cloud and have the data on premises, because it, just doesn't work and so, you have to try. To take. The, applications. Like. That in whole and move them along with the data or else you can't have a kind of a hybrid connected. Environment but from, an analytics perspective everything, from. A data Lake that they've got on-premises, today is moving into Azure high, performance computing, we're trying to look for workflows. That are, a little bit smaller. The data sets are a little smaller and try to move those in and, hole into the cloud and then, as we work on longer-term, solutions, for the seismic side of the house trying. To figure out how again we we can move all of it right so I don't, know if that kind of answers your question or provides any insight into what we're doing there. Can. I offer, a slightly different perspective so, we recently had a reorganization. And now, I'm they also in charge of Azure, storage which I know absolutely, nothing about but, I've, been coming up to speed and there's some amazing stuff have anybody heard about a veer anybody. Right. What. It's a watch this space, area. Right so what I'm going to tell you I don't, think is available right now but here's, what's on what's. Coming so, a veer, was a storage company and what they did was they said hey there's a whole bunch of NASA's out there that. Are relatively. Slow so what they did was they built a caching layer it, was all SSDs. And what you did was you had all your clients talked to it and then it would front-end all the different NASA's, incredible. Increase. So, much so that people said hey you, mask, latency. We're gonna take your device, and we're, gonna put.
It To mirror in all, of our branch. Offices, so. Now the branches, are there they talked to the local of ear device. And it talks to remote you know storage. Servers, and everybody. Thought that they had a local storage service, fantastic. So, next, stage oh well, we saying so five this caching, layer and. It back. Ends to a large capacity. And and makes up for latency why, don't I do that to, the cloud. Okay. And so that's what they do now that you can talk, to this thing and it can consume cloud, storage, where cloud storage is cheaper, and, effectively. Infinite, now. Here's where it gets interesting so. Then they said hey, we can also do this as a virtual. Appliance so, as a virtual, appliance they, get a VM, that has SSD. And then, you, talk to that and then it talks to your blob stores, and then, they're doing the work to be able to say and now, it can also you again, you're in the cloud you, talk to this virtual appliance, they, do the plumbing to talk to your local storage. Area networks now, why is this exciting, this is exciting, and used extensively, by. The, movie industry the. Movie industry their movie, is their, IP they, have you, know high degree of security around that they, want to keep that that, on-premises. So, what they do is, they. Do. Only one frame at a time up, in the cloud they, process, it in the cloud and then they write it all back local storage anyway. So the severe, the. The patterns. That you can do. On, premises. To the cloud cloud, on-premises. Is being, expanded, with, this new aveer acquisition, so watch that space we, might have some talks about it at Build here if not definitely, watch. It at ignite. Thank. You sorry I just had to get in there I'm so excited. You, mentioned that you. Had a culture, of checking, rechecking, and also, mentioned that automation. Is very key to your success, did you automate your software QA didn't great extent and if so how did you do it I, would. Say. Automating. That is one of our things to do so we're doing a little bit of that but that is one of our opportunity, areas, we. Want to have full automation, there so, we're just we have partial, I would say. Okay. We'll do two more here and then do you and, then the third one there. Hi. I also work for an oil and gas company so, I kind of understand some of the cultural baggage that can come with some of that and so, when, I think of agile transformation. A lot of people coming from a, more. Traditional culture. They're, looking for a methodology, they're looking for a process, give me the steps and I'll just do the steps right but, really agile, transformation, is more about changing, the way you think and it's. It's it's. And, people are typically resistant, to it because they. Don't know any better they've never seen anything else, and so, what, what are some, things that you can speak to about how. To help people change their mind so they're not they're not focused, on doing, a new process, and a new methodology, but, they're actually learning a new way to work. So. Some. Things that have helped, us along. Is. Getting agile coaches so. Not carrying, the full burden of that so, helping them guide you through the process especially. If you're new at it. Certainly. When I find, someone who I'm. Always so interested, okay that hurt person has heard that story ten times and today it clicked I'm always interested, in what made the brain click that day. But. I think also taking, people, and actually putting them in the process so this is what good looks like I could I can talk at them for hours, it's. Nowhere near as powerful as. Having, them sit with some of those teams so. I don't have to tell them what it looks like they see and. Feel what it looks like. And if you're just starting and, you don't have any of that take him to another company where people are running like full-on DevOps and just have them sit with them so. That they can see this is how it works. You. Know I've had people sit, with security. Teams at other companies watch, them run agile meetings all sorts of things like that just so they could get the idea of.
This. This, isn't your normal day, certainly. When we're onboarding people we do tell them all. That other stuff you do. Random. Meetings for random, projects, random, requests, for from your management, to do this do that it's like we don't do any of that here's. Where your work comes from here's. How you work it so, part. Of that is, actually. A huge benefit, to the people working agile, because it's it's a at. Its. Simplest. It's. A throughput, management. System for for human work right so. Then. Wow I don't, have all these extra demands I could actually commit, to what I need to do and then I can do it and if, somebody else will take all that extra burden. Off my shoulders so, that's the other thing is you have to take that burden away from those people if, you're free to work in an agile team I'm going to protect you from all those random one-off, requests, that, people working, in the old way it will maybe still ask of you, victory. A lot of people in, enterprises, that have been. Successful, adopting, DevOps have, talked about gene Kim's book the Phoenix project is their gateway drug, to DevOps, I was wondering whether or not you, found that a useful resource, well. Certainly when we were trying to learn we all read that and, many other books as well and. I. I'm. Sure we found a lot of ourselves, in that book so it resonated, in terms of okay I recognise, myself I I look a little bit like that. But. I think. So. It was just a way to understand. A way to recognize that this challenge is universal. In terms of. The. Problems, that we face I, think, the people who had because that's based in lean Sigma so people. Who have a background in lean Sigma who really get the factory, concept. They, really, really got it some. Other folks not so much. Because, part of our biggest challenge, is that idea, so, I literally, a factor, a floor, I'm building, a machine that, builds machines. And. So that's much, more efficient, than you building a one off and, a one off and, a one off but. Unless they worked in the process, and it's delivered, those machines, for them they really don't understand, it yeah it's one thing I found as you as you engage in, DevOps you will really it's useful to realize that there are two communities, talking, saying, similar things but really have different focus, and you want to pick your community, so the first is the startups, the internet companies, who, have no who, have nothing and then do this from the scratch, they're, great they're awesome but, it's not really the same lessons. That and and path that. A company, that has existing, assets, needs, to take to go, to embrace DevOps, and gene. Kim actually has a conference, called DevOps, Enterprise because. It's focused in on the lessons, of those, people, and it's a a bunch, of his videos, from. That conference, are on YouTube so I encourage you to take a look at those if that's your path. So. A lot of what we do as well is just go, do something, right so some. Teams like around we, have a tiger. Team that which, is an agile based team but the.
Methodology, That we're using is just build something, get something out from an MVP, perspective, right let's not sit here and talk for days around, how, to architect a TLS or let's just go and build and get something out there and iterate on it and so. That's. Been very successful. Thank. You thank, you again for sharing this morning. My. Company, has a lot of internal, infrastructure. So, for, moving to the cloud some of our system, admins of course are nervous. So. I mean how did you overcome that with teaching them to learn these skills and, overcome. That resistance to, change and then. My next question is we're very cost, Center, at our company I have, you had any real ROI. Information. To say hey this is saving, us money in the long run. So. Yes, we do have a very. Large and complicated business model, because we have savings, in a number, of different places we get savings from automation. We. Get savings from, infrastructure. Cost of avoidance, or life cycling, out and, so we not. To interrupt my how, deriving. Those metrics. Well. If, we, take. All. Right well if. I don't hit, it when I'm done and ask, me again I'll try, to hit it so. Certainly. There's known costs. So. It, is easy, to, take your infrastructure, cost cost, that with cloud cost, the, project. So that you have kind, of a bump in your spend to execute, migrations, and then, a reduction, in your infrastructure, costs, in. Your outer years so that should, not be too difficult what. Was harder for us to cost is hey, I think I'll have some savings, automation, based savings, so, we have some squishier, estimates, and we're, a conservative, company we don't like to do that so most of our estimates, are low-balled, on the infrastructure, cost so when I say infrastructure. You. Can avoid any. Data, center costs or infrastructure. Repurchase. So. If. You do if you get your timing, right you can gather up all your infrastructure, repurchase. And instead of repurchase. That's. The point where you move to the cloud and, so. And, by. Shedding your infrastructure, cost that's why we went to path. You can if you're, careful about it you can probably sell, fund your migrations. But. Certainly data center our legacy data centers were just coming they were becoming, too expensive, to, run we, just couldn't afford to run them anymore not with all the manual processes, that we had I mean we really did this as an imperative.
And, As for the as, for the server admins. Some. Of them are still nervous I would, be lying if I said they weren't however. We're. Shrinking so. We're shrinking our legacy, footprint, so. By providing a lot of training opportunities we're. Just shifting the nature of the work that we do. We. Haven't had any you know nobody's lost a job or anything like that we are Ries killing people where, we can some. People are not interested in that we have some awesome use cases somebody says I want to run this the air P infrastructure, for the next three years and then I'm gonna retire that's fantastic. Because. We, need that job done and we, need somebody to do it but, we also need people to be honest, about what they will and will not do, so. We've. Been pretty honest about what infrastructure, as code looks, like infrastructure. Management, looks, more like software, engineering, than, server administration. That. Can be a hard thing for them to hear but. We also carved, out opportunities, for, people to say this isn't even your job we're. Gonna give you some protection or relief bring, in some backfill labor to do your job give, you a chance to learn try this out see, if you can do it you, might be the new person who manages, this area if you if you are successful so we gave a lot of people who. Had the willingness, the, desire to learn we, gave them some opportunities. To see what they could do and I would say they all stepped up, you. Know I've found that when you have a big change, like this a transformational. Change. It's very these. Things are super, hard and you, know getting people through that Knothole over, the hurdle it, can be very difficult one, thing that helps if you're the person trying to make that change and, this. Would be a little controversial, but, just recognize, not. Everybody's, going to make it and. It's. Not your responsibility to, get a hundred percent of the people over, the the, threshold and you might, need to hire new talent, to, bring the company where it needs to be and. It's. Odd but once you realize that like, not everybody's gonna make it you wish that you can want to give people lots of opportunities, want, to get some stuff, etc but. It might just be that their, next play needs to be somewhere else and then it's easier, to make that transition at, least it's. Easier to to go, through the struggles. In. Making that transition. Yeah. And just one more thing to add on that it's when, you show them this, is what I need you to do. You. Can get that self selection process, going a lot faster, than, if you're talking about it and they really don't understand, it's like no I need you to learn the scripting language this, is what it looks like sit with this person and watch it you, can get those decisions, faster, of yes I want to do this and I'm interested I'll take this training I'll take the certification, or you know thanks can we talk about a different opportunity for, me. So. When, you're making a transition. From traditional. Waterfall model to safe, agile, you have all, of these roles, and responsibilities. That. People are used to and you have positions that never existed before you have released trained engineer you have different. Roles for a product owner and product manager and scrum master so, can, you describe, how, you manage, that Trent that your organizational. Jet has to change how did you manage, that organization. Change and also, how did you get senior, management, buy-in to create and fund these positions, that had never existed before in your organization. What's. A we're we're still in that transition. And. I. Laugh. Because I, mean I think people are always like what what's a release trade engineer, what are you talking about I don't even know what that is so, certainly.
There Are key leaders, or we said can you please go to this training it's really important, that you can support this, we. Also have a central, organization, within Chevron. That. Kind of carries, the agile, message and, scrum masters, and helps provide them our or, is that orgasm, mature, as we want it no it isn't but we're kind of trying, to grow that capability. But. I would say the coolest thing is, in terms of getting management, buy-in. We. Got some, we. Got enough, and. Once we started, moving and. We. Were moving faster. Than everyone, else, the. Buy-in just, came like. It's it was like a flood, please. Can, you teach my organization. I want this can you set up a team for this can I have that, because. You literally see, the power of the speed of delivery speaks, her itself and. Then you don't need much of a change management program, you just need to publish your results so. Do you now have a does. Your HR, pivot, does. Your HR organization, now have the ability to hire in into. Positions, that are that the end rolls that are directly for your safe agile process, I would, say we concentrated. Not I think, we have a couple roles that fit the safe framework, but, as part of our. Transformation. We did have our HR, organization, literally, go through all our roles, and add new ones and change, our definitions. So. That, we. Could and we even change some of our hiring practices, so, we had a very large HR, study to figure out how we would do cloud and digital transformation. And they've been a great partner in that but. You're bringing up a great point you know if you're coming to this conference by and large you're one of two categories, you, either work for a company whose business is primarily, software, or, you, work for a company whose. Primary, business is not software, but, for whom software, is becoming, an increasingly. Important, component, if you, work for one of those second. Type of companies. You. Meet you are in competition, for. Software, talent, you, need to figure out how you are going to compete against, the likes of Facebook. Microsoft. Google. Amazon. The top software. Talent, so, you need to be working with your HR department to, realize. That hey we have new career. Tracks, that are required you, need to take a look at your compensation. Are you, compensating, people. According to industry, norms, are you, giving, them the right opportunities. The right access, to for them to develop their skills I've, worked in companies doing, software, where software was not the primary business, and I've worked for companies where it was their primary business and, there, is a difference, and going forward, I think that there needs to be less and less of a difference. Any. More questions, good. Time. Aha. Okay we'll repeat the question, or. If, you stand up I can hand you this mic you just got to get in front of the speakers or it's, gonna be a bad day I will. Hold it for you okay. So it seems like at, some point there was a snowball effect where. Instead.
Of You having to go out people, started coming to you how long did that take to happen. Two. Years. Of. Some. Pretty. Slogging. Through my two years of slogging through mud is how I, would describe it. Stand. Up yeah. Yeah. What, we're transcribing, us so of it so as developers. For, the DoD Adam. And I are working. Basically from, the grassroots trying to get implement. DevOps and all. Your, pain points we're over here laughing at because they are struggles, we go through every day right now, and. I just wanted to comment on that because I know you saw us over here kind of just, laughing, and it. Was uh it, was a great talk, it's been a great conference so far so thank you. By. The way as, you embark on DevOps, you know there's somebody was asking about cost recovering, cost savings, etc, that's. All great stuff and there's a there's a there there but. Really, it is this. Business. Effectivity. That, helps get people over the hardened hurdle, for DevOps right, because, with the cost savings, it's like well is that a valid formula, and what about the hidden costs, and blah blah blah and people can balk and well any transition. There's risk etc but. Typically the folks that get through the knothole of like man we got a change is when they've had some, business. Failure, like, hey we had this project and it didn't deliver on time or we're, getting our butts handed to us in this area, as the competition, are moving faster, those, things, you know that that there's some business imperative. Those. You, find people willing to make change right. Often, failure, is the biggest motivator, for change so, you find, something where there's some dissatisfaction, we, need to make progress here then, you, find the Coalition. Of the Willing right, hey if you're not on board that's fine go do something else we want people who are in and then, you you as he says don't, go spend you know a bunch of time becoming. A 400. 400. Level expert, in DevOps, I think, reading the DevOps book definitely, worthwhile but.
Just, Do it just, like understand, the principles, we want to move fast and iterate, quickly and, do it and then learn and as you do that then start to educate, yourself and that, manufactures. Success, and then, that grows and, grows and, grows so, I go, to this DevOps conference, and basically, that story, that pattern, was told across. Industries, over and over again and then, the next year, I see the same person, just. Telling essentially, the same story. Except. This year was different because, last year they were a manager, of something, and this year the director, of something and then, the third year I come back and this person still is essentially the same story and now the vice president, of something so, you start, small grow. And, scale. Up and by way at some point when, you've had enough a sequence. Of these successes, that's, where you, can bring executive, sponsorship, in to say ok now let's figure out how, to do it on mass it. Is a mistake, it. Is a mistake, to, try and go from where you are to. Universal. DevOps, you'll, fail and it'll put back your effort by years, so, you want to start small get, some success grow. Get. A couple iterations and, then, figure out how to go, big in. My opinion. I'm, shorter. Than average so. You mentioned, that, that. You, sort of customized the agile process, and I, don't, I mean I've never talked to anybody who hasn't. But. In your experience what, are some, of the key, pieces of agile like what are some of the things that pretty, much any implementation, of, this process has to do. Well. Certainly running in the common, release train is critical. To map the dependencies, across the teams we couldn't survive without, that that's what that, dependency. Working, in the program increment, gives. Us the ability to move, stories, around move work around and hit those dependencies, and move as fast as we can. Architecture. Runway extremely. Important, we, have a process. A waterfall. Based process. Which was like we'd go architecture, review architecture. Review and so we have all these decisions, it could take two years to work through that so. When we get all that concentrated. In one week and make those decisions as fast as we can make them visible and escalate, them up that has also been extremely, important. And. Then I would say the basic nuts and bolts of the process, I mean people have to understand, stand, up and backlog. And and how to work differently. And and how the team members support them in that. I. Have. A whole. Bunch of questions but I'm trying to boil it down so which is one this. Is my favorite one I've never got an answer to it so I've been doing agile for about three years before. That it was rational, unified process. Before. That was kind, of probably deep waterfall, so. My question, is how much documentation is. Appropriate, because we prefer working software, to. A good document or thorough documentation, but. I. Get user stories which is make. It like the old system because we were transforming. Your desktop application into a web one and we, just get one line make. It like the old system and then. How. Can you derive a, good, testing strategy, from that without. Going reverse engineering, the code, who's. Supposed to do that, so. How much is mean in a year's time. No. One's gonna remember what they coded so where's the documentation, and, when. I started, we had a really good coach three years ago, his. Name was Henry Dittmer and he, said you're. Gonna be almost laughing, well that hasn't.
Really Happened yet and he. Said. Agile. Is very difficult to master and, requires. A lot more discipline, because, you don't have all these heavy-duty. Processes. So. Given, all of that how much documentation is, the right amount well, no way you've never gotten an answer it's pretty much an impossible question, yes, deliverable. Dictates, it to you. But. I would say we get it wrong most, of the time in. In, our spirit, of moving quickly I would say we have under documented, in some cases where where we need that. And. I would I would, say if. Your earlier in your journey you probably, need more and some of that's to comforts, the rest of your organization, that's used to seeing some of the stuff, but. At least if you've got it all online and, it's in our interchangeable. And you're not putting actual, like, Word documents. Or paper documents, somewhere keep. It in, the in your systems of record keep it as agile as possible, and keep it as modular, as possible. That's. Certainly the best way to go so that you can keep it fresh because, anybody. Who's written more than ten pages it's, pretty much worthless. Because. It's obsolete tomorrow. And. So I guess. My answer is modular. Systems. Where people can come back and change it very very quickly just I just. Needed to change a small piece yeah, we. Try and keep everything in one place in the user story we use JIRA. But. I actually. Do have one more question is when. You're actually writing the code I understand, that you have conversations as you go along but you have to write those conversations, down and that's something. That doesn't happen much, that's. A problem I think I. Agree. We're looking forward to some of that technology we saw yesterday we, working. We. Need a mind reader you. Buy. One of the other answers for documentation. Is to. Replace, documentation. With tests, right. Ultimately. Where you'd like to get to is a world where you say. Developers. You can make any change you want as at. When, you're done when you check it in we're gonna run these tests, and if it's green we push to production and. If it's not green we don't we back out your change and then if you something goes to production and you didn't like something that happened, then the answer is add, another, test, so, tests. Are very precise. They, say yes they say no when, someone writes something down what. Do they really mean and. As. Are we all speaking the same language I, mean literally English, and is, that your first language and, do, you write precisely, or, do you write for both tests. Are nice and precise it, doesn't do everything but. Degree, to which you can use tests, and set a documentation, you're much better off I think that's a great answer thank. You for that I just inherited some. Software and it's got 4% coverage. And I've been told it should be 75, and why isn't it there yet, from. The person you had that software for five years before me. Anyway. Thank, you. I'm an engineering. In. The oil and gas business and, I one simple question do, you get your engineers, and Chevron to adopt agile. You're. Talking about DevOps but, the. Same principles, could be adopted, by the engineering. Teams in Chevron, are you doing that, we. Are doing that in pockets, because.
We're Seeing. We're. Seeing the benefit, of that so I would say there's no broad adoption, across, all the engineering, teams but, there are some teams that do it very very well, and. Actually they understand. Probably. The. Gene Kim more. Because. They most of them had a very very, strong grounding. In lean sigma and so, once you relate it to that they. Understand, it very very well but, we don't have broad adoption, across, but. I think it's part of our digital transformation. Will be pushing out into engineering. Thanks. Lou. Yes. I after, a man, technical. Person not not. A not a part of a management side but I'm going to ask more about in the resource. So. You are. Hiring, talents right part. Of the resource of mentation of your comp
2018-05-11 08:16
