Intellyx Podcast #2: Mike Kavis – Cloud Tech is Easy, People are Hard

Intellyx co-hosts: Jason Bloomberg, Jason English.

Topics covered:

• We go way back together to being pundits in the early WS* and SOA days with Agile as a discipline, then early cloud and API-driven integration, with DevOps as a practice. Why do these megatrends become passé over time?
• As the cloud-native vendor landscape evolves, what are the biggest changes and challenges faced by companies delivering applications in the last 2-3 years?
• How do companies know how mature their organizations are on their cloud journey?
• Is Kubernetes really going to be the third wave of distributed computing, or will it be AIOps, or likely be defined as something else?
• Why is it so much easier to manage technology, than manage people?

• Mike Kavis “Architecting the Cloud” book: https://www.goodreads.com/book/show/17082288-architecting-the-cloud
• Video link: https://youtu.be/9Ejqv2A7WLc

Listen/download the Podcast on your favorite player here: https://anchor.fm/intellyx/episodes/Intellyx-ThoughtCast-2-Mike-Kavis—People-are-the-Challenge-to-Cloud-Modernization-efbfps

Watch the YouTube version here: https://youtu.be/9Ejqv2A7WLc

 

 

Transcript of the podcast:

Jason English: [00:00:59] Yeah, it is Mike, it’s great to have you on our show for the start of our Intellyx podcast season. and glad you could join us.

Mike Kavis: [00:01:10] Hey, great. To be back a different topic than a few years back, but, you know what, it’s all kind of related.

Jason English: [00:01:16] Right, you know , I don’t know if most people would know this, but we all go back to probably The mid oughts, around 2005 or so when  the whole SOA movement started up, web services, SOA, WSDL, Jason Bloomberg had published a book on the topic. I was at a young startup vendor (iTKO) just trying to publicize our capabilities for testing and you were out there —

Jason Bloomberg: [00:01:37] Catalina Marketing.

Mike Kavis: [00:01:38] Yep. Catalina marketing, yeah. I was trying to sell SOA.

Jason English: [00:01:45] So, what have you been up to since then?

Mike Kavis: [00:01:48] Oh, this thing called cloud kind of crept up in the 2007-8 timeframe I left that company, we talked about it, did a startup and learned how not to do cloud pretty good, and built a company that sold. I stayed with that company for a couple of years and just decided, you know what, I really enjoy this cloud stuff.

And I want to take all those lessons learned and, go into consulting and help people. I wrote a book along the way on it. Jason actually cleaned up my chapter on REST. So it made sense. And yeah, that was a while back 2012, I think it came up to that beginning of 2013 and it’s funny. I spent most of my life in application architecture and since that startup, because we were all app dev guys, I became ops guy and I’ve been focusing on operations ever since. And right now I’m actually my next book I’m working on is on maximizing operations. And it’s more about how to rethink the people and process side of it. Instead of the tooling side of it. So I spent a lot of time talking about like AI Ops now and SRE dev ops, all that kind of stuff.

And I, I feel like I kind of left my technology chops behind and now I’m a psychologist.

Jason English: [00:03:08] Right. I mean,

Mike Kavis: [00:03:09] Mostly for myself.

Jason English: [00:03:11] Yeah. It’s interesting how, those initial trends, we were talking about how SOA kind of became passé over time. And you know, now we’re moving through this process of, API driven integration and then DevOps and now, people start calling  it a DevOps team and they have DevOps tools.

is that a good thing you think?

Mike Kavis: [00:03:32] Well, it’s good that people are talking about DevOps. It’s the challenge. And we had the same challenge with SOA, right? So step one, buy an enterprise service bus and then sit there and stare at it and say, what’s next and then DevOps – there’s a lot of that too.

You know, you started buying all these tools and see like, DevOps equals CICD and DevOps is really a much bigger thing than that. So, you know, we get DevSecOps, Rugged DevOps, you get all these things. And I used to, I used to not like it, but then I think it was John Lewis made a good point. He says, ‘well, at least people are talking about security.’

Now at least they are engaged or involved, call it what you want. And I said, you know what? You’ve got a good point now where we’re talking about security and sprint zero, right. And early in the phase. So. Valid point, you know, call it what you will, you know, just bite my tongue and move on. And hopefully people are really, you know, it all goes back to architecture, always has, and it always will.

Right. How are we building? Are we building these things to be reliable? Are we building to be sustainable? All those types of things.

Jason Bloomberg: [00:04:32] You talked about how SOA is passé, do you think cloud is passé – or cloud, has worn out it’s welcome yet?

Mike Kavis: [00:04:38] Yeah, that’s a good question. I think the problem in the cloud space to my opinion is that many many people still see it as a data center in the sky instead of a platform for agility innovation. And there’s such a fear of lock-in, you know, people with IBM mainframes and Oracle databases are afraid of a lock in and I’m like, ah, you’re, you’re locked in. Right? So they, they go by a lot of smaller company solutions that they get locked into so they can be cloud agnostic.

So I think, people understand that cloud is a destination, but there’s different levels of maturity and understanding of the real value and the value. Isn’t so much price. I mean, there are some price advantages if you want to consolidate data centers, but it’s really about what you can get out the door.

With the smallest investment possible. Right. You know, you have IoT services, you know, you do that stuff yourself. You spend the first six months setting up infrastructure and software where now you can just go grab an API and start testing your hypothesis on these types of things.

Jason English: [00:05:46] Yeah. what would you say are some of the challenges?

I mean, you you’ve been working with  customers almost all the time now. So what are some of the biggest challenges you’re finding ? Is it mostly a mindset issue, you’re finding, or organizational?

Mike Kavis: [00:05:59] I say this all the time. So anyone who’s listening to that knows me.

I apologize for saying this again, but technology’s easy. People are hard, right? It’s the same. It’s funny. I have a deck I wrote during the SOA days called SOA and organizational change. And I could do a replace also a to cloud, to AI ops, to any of these buzz words. And the deck would just work fine.

A lot of this is, you know, if we’re going to build in the cloud cloud native the right way, you know, it’s a highly distributed environment. We’re used to a web server and app server and a big honkin database server, you know, not a distributed problem, but we take the same processes for the same tools

I was in a meeting the other day, the infrastructure people say as long as I can have all my same tools in the cloud and you know, red flag comes up, ‘uh, not all those are gonna work in the cloud’. So people know what they know they’re comfortable in their domains cloud. If you use right, can be very disruptive and very different, and getting people to change and think about the system as a whole.

And getting VPs to change a lot of times just like ‘this operations group really belongs over here in a platform team. All those are my people,’ you know, so a lot of the barriers to adoption is more people – process. Every place I go, they got smart, smart people and they can figure out the tech, they just stay at it enough, but they can’t change those, other boundaries that prevent success.

Jason Bloomberg: [00:07:23] Yeah. So, technology is easy. People are hard, but people aren’t impossible. Right? We have made progress, since the SOA days, we did make progress sort of separating concerns, focus on services as opposed to, you know, more loosely coupled perspective on the organization. and cloud is introduced organizational changes as well.

And now cloud native. Is doing the same thing and dev ops is really all about organizational and cultural change. And so it sort of stands apart from the technology in a way, except as you say, it’s still tools driven to make an excessive amount of work, but it’s still intended to be a cultural and organizational change.

So are we learning the lessons, are we getting better at this organizational part of the story? Or is it just a new generation of people come out of college and we have to reinvent the wheel each time?

Mike Kavis: [00:08:14] So. Companies who are having success or are seeing DevOps as an organizational change, a transformation companies aren’t having success or seeing DevOps as IT automation only.

So they first step one: create a DevOps team, create a new silo to solve your silo problems. You know, step two: build all this stuff and hope the developers use it, right? None of that solves any problems. But you know, I go to this annual DevOps summit that Gene Kim’s company does.

And the thing I like most about  — you’ve been there. I think I saw you last time I saw it was going to be there.  The thing I like about those shows is what I call repeat offenders. They have some companies that come back each year, so you get to see their journey. And if you go back and look at the video from four or five years ago, the problems they were trying to solve versus the problems they’re solving now.

And a lot of people were a little disappointed with the last one. Because there was a lot of talk about compliance, compliance, but that’s where these companies that are five, six years in their journey, that that’s their bottleneck that they’re solving now. So when I look at those companies, they get it, they, they invested in training.

They invested i n people, they totally blew up their operating model and They went after one bottleneck at a time, the early bottleneck was the CICD that year. It was, how do I get consistent builds? How do I get consistent environments? Then it was like, okay, testing’s a bottleneck. How do we shift testing left?

Then security’s the bottleneck, how do we shift security? Then it’s like operations, you know, how do we move Tier 1?  Or tier three support closer to the builders. And then it’s like, okay, governance, risk. so it’s just a continuous mindset that we’re going to chip away at this, and we’re going to continuously prove it, give us feedback.

And there’s a lot of companies doing amazing things there, but there are a lot of companies that aren’t.

Jason Bloomberg: [00:10:03] So in terms of the large traditional enterprises, sort of, how do you see it breaking out in Deloitte’s core customers anyway. So how many of them are really getting it and moving forward, how many of them are still stuck behind one bottleneck or another.

Mike Kavis: [00:10:18] Yeah, that’s, that’s tough because if, if you’re really doing it good, do you reach out to a consulting company? You know, if you already figured it out, do you reach out to a consulting company and we gotta figure it out? Right. So a lot of the people reaching out to us are like, ‘we don’t really know how to do this, or we’ve been doing it for two years. And it’s not working.’ So it’s hard for me  —

Jason Bloomberg: [00:10:35] What’s the point to reach out to you for the ones that figured it out?

Mike Kavis: [00:10:39] Yeah. I mean, there are some that are blazing a trail and can still use help. But a lot of the times we’re coming in, at least me I’m coming into an area that’s been trying it for awhile or are struggling.

But when I go see these presentations and when I, attend now, there’s there’s a lot of live conferences and events. Now we’re practitioners get up and speak. There’s a lot of people and wait, we’re not talking about Facebooks and Yahoos and Amazons. There’s a lot of regular companies that have pockets of people and teams.

They’re doing some amazing stuff. And even some companies that need a lot of help have a pocket of people that are doing great things. I won’t name one of the retailers, but their .com is one of the great case  studies. But if you talk to the infrastructure data center, people who aren’t related to them, They’re in a primitive state.

So it’s a mixed bag, even within companies.

Jason Bloomberg: [00:11:33] Yeah I’m trying to guess the retailer, but actually it’s several come to mind that were all in the same boat, who knows?

Mike Kavis: [00:11:38] Probably describe all of them right there. And typically the people doing customer facing stuff that require rapid deployments and quick. They’re more cloud native.

They’re more apt to go that route anyway.

Jason English: [00:11:51] You get into some of these financial companies, they have teams that are. Microsoft sized almost, right?

Mike Kavis: [00:11:57] Yeah. the amazing thing about finding it when cloud first started becoming exciting to enterprises, and I’m going to, I’m going to put that around.

2013. I think there was a re:invent where all the IAM stuff came out, all that enterprise security. Cause I know that’s when our phones started ringing and it shifted from like 50, 50 private /public to like 90% public, almost overnight. It was when that came. And, you would think that the finance companies would be the last ones because of security.

They were the first ones. I mean, there’s media and all that, but they were, they were head-first and there’s still a lot of banks today that are at nowheres, but there’s a lot of banks that not only were early, but they actually drove a lot of the decisions from the cloud providers, because they were pushing the limits.

You don’t have this, you don’t have this. And they were actually helping, driving the product.

Jason Bloomberg: [00:12:50] For certain lines of business within a big bank. So that it wasn’t core transaction processing.

Mike Kavis: [00:12:54] Now I still haven’t seen that. Right? Yeah. I mean, they want bare metal. They want low latency, you know, it’s generating billion dollar days.

We’re not gonna touch it. There’s a lot of that,  but everything else. Yeah. You want to look at your 401k, you know, all that kind of stuff. Yeah, they’re all over that stuff. And big data and analytics. That’s really where the key is to get more customer centric, to look at all this data coming in. And when a customer logs in, you know, present them with, you know, look at this, think about this, you know, those types of things.

Jason Bloomberg: [00:13:26] So last I heard you were a cloud architect, but you, it sounds like you’re doing really a much, much broader area of expertise now than  just Cloud architect.

Mike Kavis: [00:13:35] Yeah. It’s, it’s funny. Uh, well architect’s an abused term anyway. Right?

Jason Bloomberg: [00:13:40] Let’s go and use that some more.

Mike Kavis: [00:13:42] Yeah. Yeah. But really right now, like I said, I’m like part psychologist now.

I’m focusing mostly on op model redesign, new ways of thinking about operations, SRE, AIOps, those types of things. Mainly I chose to go that route because I think cloud adoption is hindered. By that side of the house. Not that there’s anything, it’s just that I think people dove into the cloud only thinking about the app and then they got the app into production and it broke and was like, well now what do we do?

Right. I got to open up a servicenow ticket back to the data center. People don’t even know I got this stuff in the cloud, so there’s a lot of focus there. And the op model is important too, because like I said, people bring their own tools, their own processes, you know, I could do a push button deploy, but then it only goes to stage and I got three months of meetings to get it the last mile.

So those are the things that, you know, people do this big ROI analysis at the beginning and it’s going to save us millions of dollars, but they never get there because they got all these cultural and process barriers and every day it takes you to get that thing in production and not shut that thing off in a data center is a day your ROI doesn’t work real well. So I used some choice words here I was going.

Yeah. It’s for children,

Jason English: [00:15:05] Right? so what do you think about the migration from just calling it the cloud as a general term to, the latest rage is the cloud-native movement. And I think a lot of people just automatically say, okay, we’re going to equate that with Kubernetes and maybe Istio now, or like they just like pick a couple pieces of it.

We call that cloud native when there’s really, I think our perspective has it basically spanning, all of your legacy technology and everything that you’re coming out with in the future, then how’s that going to deliver the —

Jason Bloomberg: [00:15:36] Abstraction that covers all of hybrid IT as well as the edge, as opposed to just being a Kubernetes story.

Mike Kavis: [00:15:43] Yeah, don’t get me going on Kubernetes now.

Jason Bloomberg: [00:15:46] Get going on Kubernetes.

Mike Kavis: [00:15:50] Well, I think Kubernetes is great. I just think nobody should care about it. It should just be running underneath the covers,

Jason Bloomberg: [00:15:56] It’s infrastructure, right?

Mike Kavis: [00:15:58] Where containers started, right? Yeah. And now you’ve got people going, building their own Kubernetes clusters. It’s plumbing at it’s max without the hardware, right? It’s all the plumbing, all this stuff that has no business value.

We’re spending a lot of cycles on when this stuff is as a service. You, you, if you use a PaaS solution, that’s all Kubernetes underneath it. And you never even have to say the word. If you’re on these cloud vendors, building, you know, natively on their platform, it’s Kubernetes as a service.

Uh, it’s great to have Kubernetes on your resume and on your LinkedIn, but it, you can go down a rat hole for months and months and months doing this stuff. We had a project, a previous company I worked at where they had a requirement where they wanted to be cloud-agnostic, whether it was the right choice or not.

So. we had to stand up an application and most of the time was getting Kubernetes set up, working, patched, continuously patched and all that. And a fraction of the time it was actually building the app. So, you know, if you don’t have to be cloud agnostic or even if you have to be cloud agnostic, I’m like use the container service on one cloud and use the container service on the other cloud.

Don’t go off and build this, you know, science project, your own platform, and now a lot of people probably don’t like that, but, um, I kind of look at it as, how am I going to get product to market as fast as possible? You know, if that’s not a priority then– okay. But most of the time that’s the priority. So,

Jason Bloomberg: [00:17:26] You know, people, when they say plumbing has no business value, it’s only true until your toilet stops working.

And all of a sudden, you know, plumbing is mission critical.

Mike Kavis: [00:17:35] Plumbing is important. The question is, do you do your own plumbing or do you hire a plumber? Right? Yeah. It comes down to that. So when a company says my plumbing is different than everyone else’s plumbing, you know, you have to go build your own. You have to ask yourself, is that true?

And second of all, if it is, why is it true? Maybe you need to change the way you think so, because that’s not the differentiator it’s important, right? Just like electricity, you know, all is important, but you know, I’m not going to wire my own house to put in a new light bulb. Right.

Jason English: [00:18:10] Yeah, definitely so any other hot topics going on right now that you, uh, you’re noticing like as far as trends or something that seems to be up on the horizon soon

Mike Kavis: [00:18:21] AI, machine learning.

Jason English: [00:18:24] Oh yeah.

Mike Kavis: [00:18:25] Everything autonomous. You name it.

Jason Bloomberg: [00:18:28] In the SOA days, there was one hot topic and now it’s great because there’s so many different hot topics. We got typecast as the SOA guys. And when SOA faded, so did we. So now at Intellyx. We’re not typecast. Because, one area goes away, like big data sort of faded away. Who cares? I just go onto something else.

Mike Kavis: [00:18:46] Yeah, almost every vendor solution now has AI on it in some form or fashion and probably 12 other buzzwords. It’s crazy. And you know, just today we were talking with a client and you know, like every client, there’s tons of tools they use for logging and monitoring. And that’s not a bad thing.

Right. Cause there’s so many different things. Every one of them now is pitching that they have AI ops. Right. So if you’re trying to implement AI ops, do you buy AI ops platform or do you just leverage. 20 different apps, but you know, it’s just crazy. And if you peel behind the covers, probably none of them have AI capabilities.

Jason Bloomberg: [00:19:25] Some, but it’s usually anomaly detection masquerading as a whole lot more than anomaly detection, but everybody can do anomaly detection. That’s pretty straightforward. It’s like a machine learning 101.

Mike Kavis: [00:19:36] Yeah. I mean, when I was doing my startup back in 2008, we were using APM tools and they were doing a pretty good job of telling us beforehand that this resource is heading down a wrong path.

That wasn’t really artificial intelligence that was like, trending. Right? But now they probably call it AI just so they can put that sticker on their product.

Jason English: [00:19:56] There’s a few different players in that space, but we definitely have known as AI ops is one. There’s hardly any just ops tools anymore, right?

Mike Kavis: [00:20:05] Yeah. So another topic that keeps coming up is resiliency, resiliency, resiliency, and in two different forms. So one, you know, there’s this big push to SRE and, and there’s all these infrastructure people trying to help make the systems more resilient.

And then there’s resiliency as in DR. If Coronavirus comes or a tornado comes. How do we, now that we kind of got good at failing over our mainframe, but now we have this hybrid cloud. How do we fail over that? And what do we do? Is it the same or is it different?

Because you know, the clouds have different regions and so and so. So there seems to be a lot of talk, at least in the last few months where I’m getting pulled into on resiliency, both from the app side and the infrastructure side. I don’t know if you guys are seeing the same thing.

Jason Bloomberg: [00:20:55] Well, yeah, I mean, it’s, it’s, it’s always been important.

It’s part of the business agility stories, part of the cloud story. And now in the, in the COVID era, it’s important, not just from an IT perspective, but increasingly from our business perspective. So business resilience has always been important, but it’s becoming more of the executive focus, you know, because business continuity and surviving the turmoil is now top of mind. So you have business resilience and then you have the combination of what’s happening on the business side with how technology is rising to the challenge and the supporting work from home and the shifting customer demands as customers purchase from home as well.

So technology resilience is becoming, especially coordinated, especially in pockets, right? Whether it’s banks supporting that PPP loans or. You know, retailers supporting demand for a toilet paper or whatever, you know, there’s these odd, you know, shifts in demand that are causing a system to go down or unemployment agencies that just aren’t able to deal with with the load.

So resilience is becoming a hot button issue because some of these pockets are not resilient. And so all of the issues are coming to life. It’s like, you know, people are waiting on hold for their unemployment for hours and hours and hours and it makes the news. And now, you know, everybody’s wondering why these companies or agencies aren’t resilient.

Mike Kavis: [00:22:28] Yeah. I think there’s two things that’s very different with this instance. One is usually it’s, a weather event or some event that happens in an area. This is global, you know, that’s one thing, but the other one is, it’s not just you this time. It’s your whole supply chain. Yeah. Everyone. So it’s at a scale that we’ve never seen before.

And you’ve seen a lot of companies had a rush to remote work. And cut a lot of corners to get all that working, but they got it working and now, you know  conferences are all online. There was one project at Deloitte not mentioning names, but where literally tens of people fly to this place and they get in the room and they do this massive deployment — that was off the table.

So it was, and that’s the way it’s been done for years, you know, client kind of mandated and it was done flawlessly online, which people started saying, well, geez,

Jason Bloomberg: [00:23:21] Well, why were we ever doing it the other way?

Mike Kavis: [00:23:23] — and we don’t need that expense and we can end the planning. Right? We can do this stuff on a dime now. Cause it’s just getting people in front of their computer as opposed to, how do we coordinate a hundred schedules to fly into this one location?

So it’s changing a lot of dynamics.

Jason English: [00:23:39] Yeah, it’s forced a change there. And it probably also caused the skills gap to come to light even more than it was before. I mean, we were kind of surprised when we go, to a KubeCon or some of these other shows and you would see, Hey, look, there’s, you know, Home Depot and CapitalOne they’re looking for SREs.

Right. if anything, I think it’s probably highlighted that quite a bit more too.

Mike Kavis: [00:24:03] Yeah. And a lot of those companies, like you mentioned, you would think they’re big and  — they’re so advanced. Both of those that you mentioned are so advanced, not just in SRE, but in big data analytics, they were early to the cloud.

You get surprised. Uh, yeah. You know, probably not the whole organization is like that, but there are pockets within these organizations. They’re just unbelievably tech savvy and moving the needle.

Jason English: [00:24:27] Great.

Jason Bloomberg: [00:24:29] Well, if this is a half hour podcast, we’re nearing the end. Right? So I guess any final words?

Mike Kavis: [00:24:36] I, you know, I go back to technology is easy, people are hard and technology is not really easy.

It’s just easier, get enough smart people they’ll figure it out. But if you really want to adopt anything and have success at any scale, meaning scale throughout the organization, you really gotta rethink operating models. You’ve got to rethink business processes. You have to kind of do system thinking it’s not just the dev team is going to do it.

It’s not just the ops team. It’s everyone. And the more we move towards higher degrees, you know, it’s kind of the DevOps story, right? The more we collaborate and work together towards the same purpose, the more we can take advantage of these technologies. There’s so much good technology out there. But it’s so hard to take advantage of it when we don’t change our ways

Jason English: [00:25:25] Well, thanks for joining us, Mike. It’s always great having you on for the first time.

Mike Kavis: [00:25:33] Yeah. The first time again. Yeah.

Announcer: [00:25:37] Thank you for listening to the Intellyx podcast. If you have any questions or ideas for future episodes, feel free to drop us an email at PR@Intellyx.com. Until next time, keep on transforming.