AWS & the Uber(izaton) of Enterprise Tech
Transcript
com. Bobby, I'm wrapping my AWS Re:Invent. I think a good portion of the 50,000 plus people expected at the show have come here. We're at the Cosmopolitan, uh, in a suite, on the ground, listening to the chatter. For those that don't recognize Bobby Allen. CTO CloudGenera. >> Mm-hmm. >> Bobby, before we get into the conversation about visibility in Cloud apps and the business outcomes of all of that. Who is CloudGenera and how do people find you guys?
>> So CloudGenera is an analytics company based out of Charlotte, North Carolina. And uh, we call ourselves almost the Consumer Reports of Cloud. So if you want to know kind of where and how to run your applications. Do I put it in Amazon, Azure, Google, On-prem? Number one, and then not just where I run it but how I run it? Do I do EC2, do I do Lambda, do I do RDS? That's what we do. And we try to do that in minutes not in months so that someone has quick answers about how to move forward.
So that's really what we do we're vendor agnostic and we want to make sure people are informed about the decisions they're making and so we often say that smart leaders start with opinion but they finish with fact. So we're the fact but we want to do that quickly so you can have some good answers. >> And I think one of the most frustrating things about whether you're talking about private cloud public cloud is getting reliable data. When I'm talking to people in my role as architect at VMware trusting the data source of what you have is one of the biggest challenges in an infrastructure.
>> That is. >> Like the executives want to make decisions but I find that one of their number one challenge is that they don't trust the unlying data provided to them to be able to make these business decisions. So I think in that theme or that path that I want to talk to you about this concept of business outcomes. And I think we talked about kind of the example of Uber. Both of us independently have used this example. And it's funny you know, Uber as we're coming off the heels of General Motor making their big announcement on their business transformation and the resulting restructuring of their business.
And they hinted to Ride-Sharing as one of the big catalyst for why they're making these decisions. And if you look at Uber, the Uber CEO was on I think CNN this morning and he said that today Uber is twice as expensive as owning your own car. So if you wanted to replace the car ownership experience with Uber today, it's twice as expensive. >> Mm-hmm. >> So I think the answer, I'm asking you this question, but I think the answer to this question is fairly obvious, why would anybody want to use Uber?
>> Yeah. So it's a great question Keith. And I think it's very analogous as we know to Cloud computing the public cloud especially. Uber cannot just be about costs. Uber is more typically about safety, about convenience, about hassle. I want to go to a different city and don't want to rent a car, certainly don't want to buy a car. Um, I want to be able to check email while I'm being transported. So the ability to multitask and do other things, not cost.
And so one of the things you and I were talking about before our time online is that, you know, looking at purchasing decisions without behavioral change is going to waste you money. And so we often say at CloudGenera looking at what you need to buy but not what you need to change will often lead to waste. If I have the same patterns with Uber that I have with my Honda Accord that I drive if I drive all the same places, I am going to lose money.
If I'm willing to drive less because maybe I'm in a more walkable area or I just don't need to leave the house as much then I can save money. But again it's not just about the dollars it's about the value. And the value is I'm not going to have accidents, I don't have to worry about insurance and maintenance and all those other things. I don't have to worry about dealing with mechanics, it's those other things. So I think to sum it up, if people are just looking at cost and not looking at the value of those other headaches that they don't have, they're missing part of the picture.
So I can totally, uh, respect the fact that the Uber CEO came out and say that, I don't know if Andy Jassy would say that. (laughs) But the reality is if I do all the same things in public cloud that I'm doing in my data center it's probably going to be the same, I'm going to lose money. Going to microservices and platforms and containers and all that, takes money to invest in that, um, so it's not just like selling a car and going to Uber you have to make an investment to go to those Uber-like services in public cloud and that's when people miss a lot of times they assume they can just cut off the data center and go to public cloud but there's an investment you got to make to consume that the right way, and there's cultural change you have to make, right?
If people just spin up stuff and leave it running for six years like they do in the data center, you're going to lose a truck-load of money. >> Right. >> There are different patterns, there's different governance you have to think about it, as you talk about a lot, that I respect, data management. Where's your data going to actually live and who is going to manage that or who can access that? So there's so many different kind of complex things, but again to net it out if you're just looking at cost and not value you're on track to lose a lot of money.
>> So let's talk about this value perspective I've talked to a CIO, this was maybe about a year and a half ago, as my role in CTO Advisor and he was telling me that they were all in on AWS they didn't do a at rationalization, they did what they did today in their data center and AWS just became an expensive outsource, a outsourcer for their enterprise data center. But this is one of the first time's I had this conversation, he said he was completely happy, even though it was more expensive, because he got value, even though he was doing the exact same thing he was doing operationally, well I wouldn't say operationally, because operationally this is where he saw the advantages.
He was doing the same thing from a technology perspective he had Vms, three tier applications, he didn't re-factor his applications, but he said Keith, the ability to software define my infrastructure makes me completely agile when the business is ready to do something I can react at a speed that I could not react inside of my data center. If I want, uh, if I want to deploy a application with three nines of availability versus a application with five nines of availability, there is no barrier to entry other than costs to do that when I embrace the cloud.
So, you know, there's value. You know, going back to our Uber example, I hate car maintenance. The fact that, you know, at some point my car comes off of warranty, or even when I was on warranty, I have to take it to the shop, uh, I live in Chicago, so I have to dust the snow off of it every morning. If I could afford to just Uber everywhere, I would. It would make my life simpler, and there's value in that. So I think we've painted two extremes.
Like the extreme of the CIO who has embraced this Uber, uh, type of bring my applications as they are and I get value out of that, to this other extreme of re-platform your every application you have and go different direction. What commonality do you see from a value perspective? >> So as we were talking Keith, there's another analogy that popped in my head and I think this is relevant for what we're seeing in enterprise, so I will confess, I do not cut my own grass anymore.
So I got to the point where I decided it was more valuable for me to spend the money for a landscaper to cut the grass because I wanted to spend more time on relationship. So when I'm traveling, I can't spend two or three hours in the yard, I want to spend time with my wife and my kids and that's where I think a lot of the CIOs and CTOs in the business are, I want to spend more time on relationships with the business units and the people who are actually doing thing that are differentiated value for the company than managing serves.
So outsourcing that job, kind of like the landscaping lets me spend more time on relationships. And I think that's what we're seeing is a transformation where the people who used to be kind of the sole provider of IT services now want to be a broker and want to be kind of a bridge between what the business wants to accomplish and how to get that done and wants to leverage those other providers like I do to cut my grass. So I think the relationships are really the piece that they see that's going to make their business special and they want to get the commoditized things outside of that and let someone else run it because it's not special.
>> So I have two questions for you along that train of thought. The first is, how do you measure success? >> So that's a great point and I think measuring success goes a lot back to what was your goal? Right? So were you trying to cut costs, right? As we've talked about this often, is short-sighted goal. In the beginning, you may have to invest money to, and change behavior to cut cost, but again cost is usually not going to be the biggest factor.
I think you need to look at, you know, am I trying to exit a data center and I trying to improve agility, am I trying to be more responsive to the business? I think establishing the goal is the only way to measure that goal. You can't succeed if you didn't establish what you're trying to do. There's a colleague at our company who says, when you don't know where you're going, any road will get you there. And that's how a lot of enterprises are doing it, they're just moving in a direction, but they never define the destination.
So defining the destination is one. I think defining your budget and your risk tolerance is another one. So what is it that we're willing to spend and what is it that we're willing to kind of, gamble, if you will. So am I willing to go a different route to that destination because it may cost more but it's more scenic. What am I trying to do, what are my goals? Is it just getting there as quickly as possible? Is it spending time with the family?
I think those goals are really important because, so I'll give you an example, if I'm a business, am I trying to deploy this app as quickly as possible or am I really trying to use this app as a learning opportunity to understand, I'm going to stub my toe or bump my head on Cloud stuff, this app is about learning, it's not about being the most efficient, it's about gathering information that we can apply later. So I think the first failure of businesses is they don't define what your goal is, they're just committing a science fair project, I wanted to do a bunch of interesting stuff, but they haven't determined what is the business value we hope to get out of this.
If you do that, I feel like you can measure it. What is the time horizon where we thing we're going to get a return on this? Who are the people that are my stakeholders? Which again, basic questions that we know coming from the consulting world, that too many people are not asking. Who are you trying to impress with this? Is this for your internal customer? Is this for the CEO? Is this for the board? Are you reacting to the fact that the board says we need to be all Cloud, so you're doing that too?
I think a lot of this is just goal setting and understanding why are we doing what we're doing? So I say if I netted it out, defining the why is probably the most important thing. If you can't do that, you cannot measure success. >> So great answer on how you measure success. And I think you walked right into my second question. Which is how do you limit the risk associated with going down this path? You've mentioned from a risk identification perspective, you need to identify, cost is a major factor in it.
There's plenty of tales now about companies that have gone to an all in Cloud model just to find out that they eat up 100% of their budget, 20% into the migration. So from a risk mitigation, whatever the risk, the risk to successfully migrating the app, to limiting customer downtime and this is where I think I get into the observability conversation. You can't measure what you don't know exists and you can't measure, you know, if you don't have outcomes, designated outcomes. And one of the things that I think as traditional enterprise architects, engineers, executives, we're used to owning the complete platform from the physical underlay in data center all the way up through the servers, to the application itself and the Cloud abstracts away a good portion of that, you know, we don't get reports from AT&T on circuit latency, jitter, etc.
We don't get the hardware telemetry data from our systems, we don't VMware optimization reports. How do we limit risk without having the same type of data that we had before? >> So let me try to tackle that question. So for one, when you're measuring whether an app fits in one venue versus another, again cost is too myopic and too small of a way to evaluate that, that's part of what our company does. So we look at things like technology fit, we look at cost, but we also look at things like SLA fit, what's the service that we are trying to deliver?
And then also compliance and privacy. So there are some things Keith, that I would argue are just a non-starter, if you're a healthcare company and you need a HIPPA complaint base to deploy your apps to, that should be one of the first things you check. And too often that happens later. They're running cost calculators, they're looking at architecture. Do you have my table stakes? The things that have to be there to make sure we're not going to jail, our name isn't in the paper, and that someone's not getting fired.
Those should be some of the initial things that we look at to determine the right execution venue in mix of services. Related to that there are venues that are okay, but some services that need to be kind of on the black list. Right? Because maybe this service is okay in this vendor, but this other service is experimental and is okay for dev test workloads, but not production work loads or real customer data. So we need to segment not just which venues, but which services within that maybe off limits, right?
Almost like, if this were cable, I'd call it a child safety rating. You can't look at all the channels and we need to say to some developers you can't use all the services. >> You know, going back to our Uber example, I love Uber, but one of the things that I don't like about Uber and Lyft, is that it's inconsistent from city to city. Like, I like UberSELECT. UberSELECT isn't available, or it's called something else in a different city. So as you talk about table stakes and then saying, okay, I'm going to go without driving, so when I go to another city, going to Hertz or Avis might be cheaper than getting the Uber and it might even be more efficient, but it's not consistent.
I want that consistent experience. So for me to even consider going to Uber full-time, I had to have the table stakes of having that feature available across my entire Uber experience. And I think that's a great example of that when you're looking at migrating to Cloud. >> Mm-hmm. A lot of people again, they just don't know what they're looking for. So if I go to Cloud, there may be tools available, but what is the question I'm trying to answer? Do I need to measure latency and is it infrastructure latency or is it application latency?
What does it look like in terms of my end customers? If I don't have a way to even define the question, internally, on prem, I'm not going to have magical tools that are going to solve that for me off prem. >> So at the end of the day, you need to know what business outcome you're looking for. >> Exactly. >> If the business outcome is that if you're a bus company and you want to reduce your total per seat cost from a mileage perspective, how does Cloud or private infrastructure or whatever you're doing, help to meet that business goal?
And how do you measure that and how do you ensure, how do you have visibility into making sure that those measurements are being met? >> Exactly. So, again, you know I talk about analogies, you do a lot too. Avatar comes to mind. Right? James Cameron had a vision of Avatar and he held that vision back because the technology wasn't ready. In the enterprise, people need to be able to do the same thing. We want a certain level of availability, or monitoring, or management that may not be possible.
And so that doesn't mean put everything in Cloud or nothing in Cloud, but the things that kind of fall above that threshold, maybe we hold back as opposed to trying to jam in a square peg in a round hole. We have to have a vision of what we're trying to do and kind of how we can attack that. Again the challenge is people, they're just expecting Cloud to be a magic bullet key. There are some great tools out there but that's like expecting that a blender is going to make a three layer cake for me, it's a tool, but I still have to have a recipe in mind that I'm trying to accomplish in the end.
And I think a lot of us in enterprise don't understand what we're trying to cook, we're just getting in the kitchen throwing around pots and pans. >> Yeah, you just made a mental note for me to return my Ninja blender because now that I know it won't make the three layer cake for me, I'm a little bit disappointed. >> There you go. >> So we've come to the end of our time. It's been as usual great conversation. My wife said, "Oh your last interview is with Bobby?
" >> Okay. >> So we can get together and we can talk quite some time. I hope you've enjoyed this episode of The CTO Dose at AWS re:Invent 2018. Bobby, you're on Twitter, your Twitter handle? >> Ballencharlotte. Ballin' in Charlotte is me. >> I can guess why that is. And I'm @CTOAdvisor on the Twitters. com, subscribe to the YouTube channel, like the video, I'm putting a little bit more effort into YouTube. Talk to you next CTO Dose.
