SERVAAS VERBIEST (SV): Welcome to IT Availability Now, the show that tells stories of business resilience from the people who keep the digital world available.
I'm your host Servaas Verbiest, and today, I'm joined by James Mora, Vice President of Engineering and Operations for Sungard AS. We're going to be discussing what businesses must consider when developing a multi-cloud strategy.
James, thanks for joining us on the show today.
JAMES MORA (JM): Absolutely.
(SV): So, multi-cloud tends to come up pretty consistently in any conversation that I'm in, and at times, can be considered the de facto standard among enterprises.
Flexera puts out a great report every year that I reference pretty frequently and they've stated that almost 89% of businesses utilize multi-cloud, but within the single-digit percentages, you still have people using one platform or even using a private cloud platform. Which isn't bad, but it kind of asks the question, how do I get there? What should I consider if I want to make that transition? Or is the track that I'm going on using one cloud not the way to go or is it?
So when companies are taking that single cloud approach, but want to make the jump to go multi-cloud, where should they begin? How should they determine what makes the most sense for the organization?
(JM): So it really starts off, Servaas, with the question: why do you want to make the jump? And working the whole problem through. Do you need to make the jump for competitive innovation reasons? That is, do you need to increase the velocity of your development cycle? Be able to push out development faster? Are you looking at cost implications? How do you outsource more of your care and feeding of your infrastructure and really leverage economics at the hyper scalars? So it starts with that question and then really understanding the characteristics of your workloads. That is, the applications themselves. And then, what is the best cloud suited for those workloads? And then how are you going to actually do the migrations themselves from on-prem to the cloud?
So it's understanding your workloads, understanding what cloud matches those workloads and then working through the migration technologies.
(SV): And you know, it's something that's always very simple to talk about, right? And if you have a lot of knowledge on this kind of project and this kind of initiative, you can cleanly illustrate like you did that transition path, but what does it look like in practice? Can you give us an example that maybe unpacks that a little more?
(JM): Sure. Well, I think the simple scenario is probably how you're operating today in a single cloud on-prem environment. And that is, you're really operating in a tightly integrated ecosystem - your application, your infrastructure, your skills are all well bound, understood. Once you start moving to a multi-cloud scenario, you've really introduced levels of complexity. And that is, your application architecture needs to be able to operate across clouds. That is, you know, whether it's between on-prem and public cloud or between public clouds itself. So web tier, database, caching - all of those where you don't have those problems if you're operating just within your data center - you now have to think through the architecture there, the infrastructure itself. Now, you may be understanding your skills, the infrastructure that you’re using on-prem. If you've moved to AWS, Azure, Google Cloud, now there's levels of understanding of their infrastructure as well. So not only the developers now need to understand how to actually move their software to the infrastructure, but now how are you going to monitor that? And how are you going to manage your application performance and customer experience to make sure you maintain a consistent customer experience across really a much more complex scenario?
(SV): You know, I like some of the things you highlighted there. But it also kind of illustrates one of the biggest challenges that I've seen, where everybody's excited to talk about understanding the workloads and the technology and the migration process. They put a lot of focus there, but what they sometimes overlook, and it can be disastrous, is supporting and maintaining these environments once you've made this transition.
So what have you seen if people don't take that into consideration and they just kind of jump in and try to figure it out as they go?
(JM): Well really, if you don't fundamentally understand what workloads and why you're moving to the cloud, then you're going to impact the customer experience ultimately, and you're probably going to not be doing it. So for example, two classic examples, really, of why you may be moving is if you have really unpredictable types of workloads. If you don't really understand that or how that is based on marketing campaigns or time of day or seasonality, how you're going to basically move your web applications to the cloud can be extremely challenging.
And secondly, really from a reliability standpoint, if you're moving across a multi-cloud environment, how are you going to know if you're having difficulty at one data center or if you need to shift your workload to another data center? If you haven't thought through all those monitoring implications, then you're going to probably be experiencing a lot more application outages and poor customer experience than you had when it was just on-prem.
(SV): And I like how you refer to it as a data center because typically people get wrapped up in that marketing buzzword that’s cloud. But the reality of the situation is these workloads are residing in hardware that sit in data centers that are controlled by other entities, right? And if you take a step back, and you kind of remove the mysticism associated with the marketing buzz and focus on, to your point, the key criteria associated to running the application and running the business based on the requirements of the end user, you've got a higher rate of success.
Now, I know we've talked about a lot of the concepts associated to it and some key criteria and examples, but how do you plan for this? What kind of steps do you think a business or organization should take to ensure they’re properly maintaining and supporting a multi-cloud environment?
(JM): Sure. Well, because you've moved to a multi-cloud environment, you've introduced a certain set of problems now that your plan needs to address and one, again, is your application architecture itself. So you need to work through how your application - caching, database, web tier - are all going to operate in a multi-cloud environment, how they're going to maintain session state across that so the customer experience is not impacted. So you have to work through your application architecture.
With your DevOps organization, which is working with your development organization who’s deploying software, but is tightly integrated and has an understanding of the infrastructure, you've essentially created a barrier between the cloud provider and your organization. So there's somewhat of a gap now that you have to address. And that is, how are you going to close the gap between your development organization, which typically has a holistic understanding of the infrastructure and the software stack. Now that’s separated so you don't necessarily know what's going on in the infrastructure world, that's kind of someone else doing that. You know what's happening in the software. So you have to really integrate so when you're doing software delivery, you know what is happening on the infrastructure side and if there's something happening on the infrastructure side, you know what's happening on the application side, and the monitoring, as well.
So now, where you had a single pane of glass and been able to monitor, that’s become much more complicated. So you still need to have a single dashboard where you're able to look at your whole performance for the customer experience holistically. But now that has to get tied into all the different infrastructure. And not just monitoring the performance of the application, but now, because you're flexing virtual machine instances up and down, you really have to have a good and transparent understanding of your cost. If all these servers are sitting on-prem and they're sitting there not being used, that's one thing. But now if you have a bunch of EC2 instances that are sitting in AWS that are not being used, that's different. You're paying for stuff that your application isn’t using. So you have to have really a transparent understanding of that across all the clouds, so that you're just using what you're consuming.
On the skill set side internally now, like I said, you've introduced complexity. Where your team only needed to know the skills needed for the infrastructure they had on-prem, now they need to be Azure experts and AWS experts and Google Cloud experts, VMware experts. So part of that planning is there are opportunities, like with all of these, to work with organizations like Sungard who are managed service providers, who can really fill that gap so the development organizations, the IT organizations can just focus on the workloads. Service providers really then fill the gap with understanding, having the skill sets for all the different clouds and tethering that in. Otherwise, your organization's really going to now be required to make sure that your staff has certifications for all of your DevOps organizations that you maintain and for every cloud that you are operating under. Your actual skills have become much more complicated in terms of you having to maintain a wider skill set. And again, on the applications itself, the reliability of the application has become more complicated because now my application is operating in different clouds.
(SV): At the risk of potentially giving away the secret sauce - and this is one of the reasons I'm excited to have you on our show today - you work in an environment where that's what we do, right? We provide optimization, maintenance, design help and cost controls for these multi-cloud deployments.
What are some of the things that you consider and that you have to put into action as you plan out how you're going to have your team support these initiatives that we're going to be doing on behalf of customers? Nothing too crazy, but just a high level so they can understand what really goes into maintaining this kind of organization?
(JM): Sure. One, first of all, is working through the application architecture that's on-prem and how that’s going to move, whether it's going to move to a single cloud or multi-cloud environment. So you're going to need to work with some experts to help you to reimagine what that architecture needs to be.
Secondly, in your DevOps organization, the provider’s organization really needs to be an extension of your DevOps organization. You can't have a barrier between something happening in a data center and you have no idea what's going on. You have to still be able to look at things holistically. So that is, the infrastructure to the software stack and that it's all tied together. It's very important that not only the team members are seen and the DevOps team is seen as an extension of your team, but also the monitoring and management tools. Ideally you want the infrastructure and the application - the tools that are used to do that - be tied completely together so that you could see the whole thing holistically just like if it was on-prem just as a managed service provider can. They can see that they've made a change to patching that impacted the application performance. So now the managed service provider knows that it can work proactively to resolve the issue because a new software deployment actually impacted things negatively, you can see the performance of it and then work in unison with the software team.
(SV): Yeah. James, I really want to thank you for being on the show today because you provided a lot of fantastic insight.
(JM): Absolutely. It's my pleasure, Servaas.
(SV): And I gotta tell you, this is great for our listeners because you really emphasized focusing on the why first and then the what. Really trying to maintain focus on the user and the customer experience balancing the DevOps operation with the developers and development that really drives an organization and most importantly, having the means to holistically review data that gives you insight on what’s going on across these vast landscapes and having a unified operations plan. If you don't have those things, I don't know how you would support a multi-cloud strategy to begin with. If you're listening and you don't have those things, you really should work to integrate them. Things will get a lot easier and you're going to see better results as you start to incorporate these cloud platforms.
James Mora is the Vice President of Engineering and Operations at Sungard AS.
You can find the show notes for this episode at SungardAS.com/ITAvailabilityNow. Please subscribe to the show on your podcast platform of choice to get new episodes as soon as they’re available.
IT Availability now is a production of Sungard Availability Services.
I’m your host Servaas Verbiest, and until next time, stay available.