10 on Tech Episode 007 – To the Cloud and Back Again with David Klee

Share with your friends


On this episode of 10 on Tech, David Klee (@kleegeek) joins ActualTech Media to discuss the difficulties with moving to the cloud. David’s professional services organization, Heraflux Technologies, helps customers evaluate and perform things like a cloud migration, and David has a unique perspective on what makes it work or not work, and what makes it worth it or not worth it.

David recently co-wrote The Gorilla Guide to Modern Storage Strategies for SQL Server – get a free copy here!

As always, the transcript is available below the player. Don’t forget to subscribe to the show on iTunes, Google Play, or Stitcher!

Show Transcript
James Green: Hello and welcome to another episode of 10 on Tech. I’m your host James Green from ActualTech Media, and my guest today is David Klee. You can find David on Twitter at @kleegeek. David and I are going to talk today a little bit about the cloud. I’ve been doing some talking with David just recently about some of the customers that he works with. David does a lot of consulting in the field at his company Heraflux, and he sees some really interesting stuff with regard to how customers are attempting to use the cloud. Where they’re having success, where they’re not. David and I have also done some work together in the past with some ActualTech Media stuff. David wrote a book with us, and I’ll put a link to that in the show notes. It’s a really great book on SQL server and flash storage. It’s a great read, so make sure you check that out.
  Before we get started, David, welcome to show. Anything I missed about you?
David Klee: No. With an intro like that, that’s incredible.
James Green: Here’s what we’re going to talk about today. Public cloud is a really exciting possibility for most organizations. They look at it and they say, “I can see how I could leverage this to get more agile to save more money, to operationally simplify.” It turns out that in some cases it just doesn’t work out that way, or, they try it and it’s too complicated or it’s too expensive. I think the biggest thing that really makes a difference is what you’re trying to accomplish. For some use cases, it’s far and away the best option to put all your stuff up in AWS or Azure or another cloud provider. You’re going to get the benefit of their economies of scale. You’re going to get their operational efficiency benefits. It’s going to be way better. Then there’s some workloads where, just based on the performance requirements or the amount of data you’re going to transfer, or something, it’s really not worth it.
  That’s what we want to talk about today. David, you gave me some really interesting examples before we started talking about some of the experiences your clients have had with trying to do that. What have you seen over the last maybe year or two years with regard to customers of yours trying to either evaluate public cloud and decide if it’s going to work for them, or if they actually did, what have their experiences been like and what are they finding works and doesn’t work?
David Klee: That’s an awesome question. You’re absolutely right. The public cloud is a phenomenal resource for all kinds of purposes. We see a tremendous amount of excitement around it. Us personally, we were building a couple of tools. It’s designed cloud first. When you look at the needs of some of the organizations that we manage, sometimes public cloud is not the right use case for it. At least not yet. I firmly believe that every single day the public cloud gets better and better and better. There’s a lot of different platforms out there for a lot of different capabilities and features. When you start digging into it, some of these workloads, especially a lot of the weird stuff that we manage, they’re very complex, they’re very resource-intensive, they move a lot of data around. Sometimes the economies of scale, because the cloud is built for scale-out and a lot of these systems are scale-up by design, by nature. A lot of times it doesn’t make sense to go public cloud for some of these application stacks. At least not yet.
  One example, the company we’re going to be working with soon. They moved everything to the public cloud about two, two and a half years ago. Over time they realized that maybe it wasn’t the right move for them. The availability wasn’t there with a lot of the areas. The security requirements were there. They were able to get a good platform for that, but the performance wasn’t there. In order to try to get a lot of the performance that their growing business demanded, they just couldn’t get it, or, that the architecture was so complicated versus what they were already doing on-prem, that they ended up moving everything back about three months ago. Now things are faster, they’re more highly available at every tier not just production, and it’s cheaper.
  I know that they’re looking at public cloud down the road, but for right now, that particular workload with how much data they move and how fast they’re growing, it’s just incredible.
James Green: I don’t personally have a ton of experience with doing the in-depth ROI analysis of whether it’s going to make sense for a company to do that like you do, but I have a suspicion that one of the things that makes it less economical to move workloads to the public cloud, and you touched on this a little bit, is the scale-up sort of workloads. As you said, in my experience, most public cloud platforms are built to support an architecture that is inherently scale-out. Some of those systems that – I know you especially do a lot of work with SQL server and Oracle databases – when those things get pretty large, I suspect that those are some of the tough things to host successfully in the public cloud. Is that the case?
David Klee: That is the case. It’s not just the databases, although they’re the bulk of it from what we see. Then again, we’ve had a skewed perspective. Think about it this way. Public cloud is designed for scale-out. If you’re designing an app for the public cloud, you build that scale-out architecture into it from day one. No problem. When you look at the, and I would claim the vast majority of business-critical applications today, I’m not going to name any names because I don’t want to call out the offenders, but when you look at these mission-critical apps, these are monolithic designs that have evolved over the last 10, 20 years? In a lot of cases, they’re built for scale-up because they don’t actually have the capacity to scale out.
  You know, a lot of these things are 5, 10, 20 terabyte single databases. A lot of these application servers require very large footprints in order to manage just the basics of them. You try to put that in the cloud, there is just no way to make that cost effective versus what you can do on-prem. At least not today.
James Green: I think it’s really interesting that we have seen examples of small businesses that try it and they find out “for my workload that’s just not right.” We’ve also seen the big guys figure out the same thing. If I’m not mistaken, Dropbox just completed a huge migration out of the public cloud and into their own data centers, because they found that they were getting to the point where they could do it cheaper and better and faster themselves. I’m curious what your thoughts are around the future of that dynamic? You said that the cloud is getting better every day, and I believe that too. What’s it going to take to make it so that more people are moving in that direction? We saw in the very beginning of the cloud era, it was cloud cloud cloud, we have to go to the cloud, and people moved stuff. Now, at this point what we’re seeing is, some people are still in that phase. Some people are already jaded and they’re unhappy with their experience and they’re going, “We got to get out of here, we’re coming back.”
  What’s it going to take for the public cloud providers to be a better option than on-premises today?
David Klee: I think its going to take time. The jadedness that you mentioned. Step back 10 years and apply it to virtualization instead of cloud. Same exact thing. In a lot of workloads the platform wasn’t really ready, or you have to do a lot of craziness, a lot of silly architecture to really get it there. Today, virtualization is basically an assumption, and it’s going to work. There are very few workloads, I can count on one hand the number of workloads in the last five years that they said, “Don’t virtualize.” You know, I look at the public cloud as getting better and better. The cost has to come down, the scale needs to go up, the speed needs to go up, the availability needs to go up. The availability and performance needs to have this work for traditional apps and not just cloud-ready apps.
  It’s getting better every day, you know. New features are coming out all over the place. New storage platforms are coming out for cloud-readiness, but it’s going to take time.
James Green: There’s obviously unique requirements, especially in the enterprise, especially around security and compliance, some of that kind of stuff that may change what people are planning to do. Is it going to be that hosting your resources, your workloads in a public cloud somewhere is an assumption just like virtualization is an assumption today? Is that how you see it going?
David Klee: I feel down the road, most likely. Again, you know the future is always hard to predict of course. I feel like down the road, the futures of the platform from a security standpoint, the public cloud right now, in most shops that I’ve been in, built right, the public cloud is more secure than their environment. Hands down. It didn’t used to be that way, but now you can just check a box and now you’ve got these different compliance levels that are really hard on-prem. When you look at it that way …
James Green

James is a Partner at ActualTech Media and writes, speaks, and consults on Enterprise IT. He has worked in the IT industry as an administrator, architect, and consultant, and has also published numerous articles, whitepapers, and books. James is a 2014 - 2016 vExpert and VCAP-DCD/DCA. Follow James on Twitter

No Comments

Post A Comment

Web Analytics