One of the major considerations when modernizing applications is how and where they’re going to be hosted—what we call landing zones. Today, you have a wide variety of options that includes, at least, some combination of on-prem, public cloud(s), Kubernetes, VMs, PaaS, and bare metal. Because of the dynamic nature of applications and the complexities of enterprise IT budgets, choosing is rarely as simple as just identifying the least expensive option.
In this episode of our Cloud & Culture podcast, VMware multi-cloud architect Adam Bohle joins returning guest Felicia Schwartz (of VMware Tanzu Labs) to discuss how organizations make these decisions. Beyond the purely technological aspects, data gravity, existing commercial relationships, and vendor lock-in all come into play, as well.
Here are some highlights from a very insightful discussion, but you’ll want to listen to the whole thing for all the cloud-packed goodness
Data gravity is real
Adam Bohle: “There's a lot of attraction from public cloud providers to … ‘Bring your systems of record, bring your critical business applications, bring the databases and the data warehouses that are core to running your business. Bring them to our cloud platform, because we can run it here and we'll make it cost effective, and we'll take away all that operational concern and maintenance of underlying infrastructure that you have. … And by the way, when you're looking at your modern applications or you're looking at implementing IoT solutions, we’ll suck all that data into our public cloud solution, as well. And .... you're using some, some productivity tool that's in our cloud.’
“By now, you've got kind of like a convergence of data, of users, they’re all in one place—it makes sense to bring the applications to that location, as well. There are obviously solutions for caching certain data workloads and so forth, but ultimately you're dealing with a law-of-physics issue at the end of the day. Does it make sense to run certain application workloads potentially hundreds of miles away from another cloud provider that's hosting the data and so forth?
“So I think that customers need to think just as much, if not more, around what is the strategy for their data going forward.Because that's ultimately going to pull in their application migration strategy, as well.”
Prioritize ‘ease of consumption’ to avoid shadow IT
Adam Bohle: "There'll [always] be something that happened ‘over here’ because of some development team that went off and created something. And, of course, we used to refer to that as ‘shadow IT.’ Shadow IT doesn't seem to be a term that I hear too much anymore; it just seems to be, ‘Well that's happened and there's some stuff over there and that's now running in production.’
“So, I think there's an element where just that ease of consumption needs to be there. Because in my experience working with developers and so forth, the path of least resistance, from an infrastructure perspective, always takes precedence if you can just hit the ‘easy’ button and get what you want. And that's why so much money is spent with public clouds —because you can just ‘next,’ ‘next,’ ‘finish’ and—boom!—I've got a PaaS solution. Boom! I've got my web hosting platform. I've got everything that I need, and I managed to do it without having to exit my tool chain.
“So I think that path of least resistance fits into play. And I think from an infrastructure provisioning and operational perspective, we need to be able to offer solutions to customers that just mean the path of least resistance is utilizing the technology tool chain that ultimately gives them everything they need from a business and IT strategic direction perspective.”
An audio excerpt of Adam Bohle discussing the difficulty of migrating many apps at once.
What’s your cloud exit strategy?
Adam Bohle: “One thing that I see quite common, or becoming more and more and more common is, ‘OK, I will consume workloads, or run my workloads, in this particular cloud provider, but what's my exit strategy? What happens if I need to get out of that cloud provider? I may not have a reason at the moment, but who's to say what will happen in 5, 10, 15 years time. Do I have a plan to get out of that cloud provider, or am I consuming services, or have my development and infrastructure team consumed services, that mean I'm ultimately locked into that location?’
“And that, as well, comes back to that whole data locality piece. If I've put all of my data in one cloud provider, that's a point of lock-in. If I've consumed some particular PaaS service, that's only native to that cloud provider, I'm locked in there. If I'm consuming other services that are only available there, that's an anchor point for all of these workloads.
“So I think a lot of customers—and perhaps I have a bias to this because there's a number of financial services customers that I work with—they're looking at, ‘How can I consume this infrastructure? How can I get the best out of this infrastructure—create those paths of least resistance to make sure that I have the agility, and I'm able to get the applications provisioned and running and performing best for my business—but also be able to get out of that cloud provider or move to another cloud provider if I need to?’”
Mainframes: Not going anywhere, and not always the problem
Felicia Schwartz: “From an application perspective ... the mainframe, COBOL, these old technologies were very hard. … Sometimes, [organizations will] just outsource their mainframes to a company, and then they could get rid of that infrastructure.That tends to be a strategy if these applications need to keep running, but they're not necessarily changing and [don’t] need to be modified similar to what was needed when COVID hit [with] the unemployment systems.
“There are other approaches that over the past few years have come into play. I know we've done a lot of work with customers on, ‘How do you quickly iterate on these so that it may take time to shut them down, but you could get the value where you need it really fast?’ So, decomposing them in an iterative manner where you have an anti-corruption layer between the legacy mainframe or AS/400 apps, and what's really needed.
“In the unemployment systems, a lot of it was just, ‘How do we get enough users online?’ So it wasn't that … the whole unemployment systems were shut down and needed to be totally rearchitected Day 1, components of it did. And those are things where you could say, ‘Hey, I'm going to leave the things that are OK and working fine, but I'm going to modernize quickly and move to the cloud—a secure cloud platform—the parts that need to change. ... We're past the urgency for that, but we're seeing that with all of our customers who have these old mainframe apps, that it's really hard to find people who know the technologies anymore, but they want to take advantage of the cloud.
“I think we're going to see this more and more over the next few years because companies that have these just need a quicker path than to say, ‘I need 3 years or 4 years to rewrite this in its entirety.’”
Learn more about application modernization
About the AuthorMore Content by Derrick Harris