A Podcast Series Dedicated to Mainframe Modernization
Transcript
00:00:00 Becky
Hello and welcome to this episode of Heirloom Modcast, a podcast all about mainframe modernization, I’m Becky Etheridge, the Chief Partner Officer for Heirloom Computing.
So in today’s episode, it’s my pleasure to welcome you to this short review of ISG’s most recent Provider Lens report on software solutions for mainframe application modernization.
00:00:25 Becky
But before we get started, let me very quickly introduce you to Heirloom Computing.
We are a software company that partners with clients and global system integrators to migrate mainframe workloads to modern stacks that are service-agnostic for execution on the cloud, on-premise, or on the mainframe.
00:00:46 Becky
And we are incredibly proud to be positioned in the ISG 2024 Provider Lens report as the highest-ranking multi-cloud provider of mainframe migration software and as ISG states in their report, Heirloom is a single platform that delivers both replatforming and refactoring outcomes simultaneously.
And that means you get to take complete control of your mainframe transformation journey, including how and where to deliver it.
00:01:17 Becky
I’m delighted that we are joined today by Pedro Maschio, ISG’s distinguished lead author for the report, and Gary Crook, Heirloom’s founder and CEO. So it’s my very great pleasure to hand over to you, Pedro, to give us an overview of the key findings from ISG’s latest report.
00:01:36 Pedro
Thank you, Becky. It is a great pleasure to be here with you today. Thank you for inviting me for this podcast.
This is the 4th year that we did the mainframe modernization study, and Heirloom has been named a leader for three consecutive years, so congratulations for the company in the 1st place.
What we see in these four years doing this study is that more companies are considering migrating off the mainframe into the cloud, because mainly because of cost. So the main driver is cost and last year with the economy uncertainty and crisis and reduction in IT budgets, more companies consider start doing mainframe. So we see an acceleration in mainframe migration to cloud.
00:02:28 Pedro
And perhaps that’s the right timing for companies to consider doing that. We also see more competitors in the market that meaning the market is attractive for more companies to participate to more service providers. And also we see acquisitions going on for the last two, three years.
There are many acquisitions and that meaning that large enterprises, large providers are also looking into the market with good eyes and good interest in the just accelerating the mainframe migration.
00:03:03 Pedro
Our survey shows that about 50% of the clients, mainframe clients, they plan to migrate off the mainframe in the next five years. The other half, almost half, plan to do in 10 years. There’s a very, very small number of companies that say I don’t have plans right now, but it’s really about planning and having an answer.
I think today companies should not, you cannot go back to your CEO and say we don’t have a plan. So, you should have a plan of what you’re going to do with the mainframe.
The other driver besides cost is the cloud scalability, so it allows you to grow in small pieces so you don’t have to buy a new machine and new mainframe core or storage, so cloud you can grow faster, easier and cheaper and also the new demands around AI, Gen AI, analytics. How do you integrate the new applications that everything now is written for the cloud, it goes in the cloud first.
So how do we integrate? Because there’s a lot of company data residing inside the mainframes and then you need integration. You need to get value out of that data and the other driver is also that we have new compliance requirements, new demands country by country.
00:04:31 Pedro
So the… and one of them is that what you do with your mainframe and not only the mainframe but the data center because of energy consumption, carbon neutrality, things like that. So these are side effects of migrating to cloud that you can resolve and then really modernization now you have a greater view into the future and more flexibility.
00:04:58 Pedro
And then I think, that we can have some questions for Heirloom and Gary, perhaps you can address what Heirloom can do to help clients in the in making this transition. What are the solutions you see for mainframe modernization?
00:05:18 Gary
Yeah. Thanks, Pedro for the summary and the insights. I know you’ve been covering this space now for a few years, very high quality analysis. So yeah, certainly happy to be here today and answer some of these questions.
So I think, look, when we’re migrating complex workloads such as those on the mainframe, there are many risk considerations. And I broadly categorize these as technical, operational and cultural.
00:05:52 Gary
So you know when you’re when you think about it, broadly, when you’re retaining your existing code, that’s something that’s commonly associated with replatforming solutions, that can be a great way to minimize some of the short term risks associated with migration. But what it doesn’t do, however is move the modernization needle.
So that, that kind of means that the other risks that are out there, you know such as the agility and the code base, those issues are not addressed, right, and that’s another issue that that’s, you know alongside, you know, the impending kind of skills issue as well.
00:06:38 Gary
So, my experience being in the mainframe modernization arena now for over 30 years, most clients know that to address the skills and agility issues they need to get beyond the constraints of their mainframe application ecosystems. And that’s why you know several clients, many clients are looking to refactoring solutions to replace the original application code, usually with Java, sometimes C#.
But the problem is that discarding your existing code base is a huge leap. If you were to conduct a careful risk assessment across those main categories I mentioned earlier, technical, operational, and cultural, that’s going to yield a large number of red flags, not least of which is, you know, the high cost of a high risk multi-year project.
00:07:37 Gary
So we believe that the right approach, as Becky mentioned earlier is one that allows you to control the pace of that transformation, one that allows you to retain your original code alongside functionally equivalent object oriented Java source code, and of course, that’s exactly what Heirloom does, and it’s this unique core differentiation that puts our clients in control of their modernization journey.
So again, you know, being in control of how and where to migrate is the most powerful position you can put yourself in to eliminate the risks with, and you know, let’s face it, what are inherently complex transformation projects.
00:08:27 Pedro
I think there’s a lot of value business value that’s written in COBOL, so it’s not really about ignoring the applications, we need to do something to maintain the applications, but we see there in the company saying, there’s a strong focus in 2023, we noticed that a strong focus on cost reductions.
Clients are prioritizing projects that can deliver benefits quickly and sometimes they look at the mainframes and they perhaps see this is complex, but how can Heirloom address that initiative? I mean how long it takes to do modernization quickly without losing business performance and business value that you have in COBOL applications and other languages, how do you address the demand for quick transformations?
00:09:22 Gary
Yeah, OK, thanks. So I think that there’s kind of two aspects to this, right. There’s the, there’s the cost structure and then there’s the speed of delivery.
On the cost side of things, I think that you know the shortest, most impactful answer I can give you is what our clients themselves have reported, and basically what they tell us is that having done the migration, they’re able to cut their OpEx by at least 60% and sometimes as high as 85%.
I know that those numbers may seem unreal, but again, in my defense, this is what is reported to us by our clients. You know, I think also anyone with a mainframe will be very familiar with the increasing cost structures, you know, especially from their primary software vendors.
00:10:17 Gary
In fact, you know, one of the common, but perhaps unusual characteristics of a successful migration is that the savings are frequently higher than the original business plan. I’m not sure exactly why that happens, but you know, I strongly suspect that most mainframe shops do not have a completely clear view of their operating cost.
That, you know, more broadly, is actually a theme in these types of projects. You know, one of iterative discovery, that occurs across many different aspects of a migration project. These are systems that have existed without intense examination for decades, so perhaps it’s not a huge revelation that there are a lot of unknowns in, in the attic. For the other point you raised, you know, essentially the prioritization of workloads with the fastest ROI. I think there are kind of two main considerations when thinking about this.
00:11:20 Gary
The first is the approach. So again, if we look at the market, you know we have replatforming, and we have refactoring. That’s the two dominant patterns.
So, with replatforming, in our experience, that delivers results, you know, much faster than refactoring. Roughly speaking, I’d say, you know, at least three to five times faster.
And that’s basically a consequence of the core underlying technology used, which is, you know, typically compiler-based solutions, you know, which is basically the same science that exists on the mainframe today.
00:11:57 Gary
Whereas refactoring solutions, you know, use more complex analysis to get a kind of abstraction of models which are then used for code generation. Now that sounds you know cool, but it’s actually very imprecise.
So for context, again, Heirloom delivers both outcomes simultaneously, but at replatforming speed and we do that with transpilers, which is a type of compiler that emits source code.
00:12:33 Gary
So the technical approach is kind of like the first aspect of it. The second main consideration I’d consider would be the type of workload.
So for the fastest returns, I’d say clients should adopt, you know, a batch first strategy. There’s a few reasons for that. Batch typically consumes the most resources, and it’s consequently you know, more expensive than the online processing side.
It’s also relatively easy to isolate, and what I mean by that is it’s actually possible with Heirloom, at least, to execute batch processing on the cloud and then to push the results and the data extracts from that batch processing back to the mainframe for downstream processing. So that’s a couple of the main considerations I would say in terms of identifying the fastest ROI.
00:13:36 Pedro
And there’s so many companies, there is still some hesitation of whether or not should I mess up with this workload site? I mean, exactly what you said, they don’t have documentation or nobody knows how it’s working and then it may be a bit complex to migrate this application successfully to a modern application stack besides that is a procedural programming and then we have a new languages with object orientation.
What are the challenges customers face, and especially why does hesitation, why companies are afraid of addressing the transformation?
00:14:25 Gary
Well, first off, I think that you know that hesitation and you know you might even want to describe it as skepticism, I think that’s a great starting place. I mean, I think when we consider you know, the mainframe in terms of its history in terms of the resiliency that it’s provided, you know, it has been the foundation for transacting business for decades. So I think that skepticism is actually very, healthy.
00:14:51 Gary
So I think that kind of leads into the realization that, you know, one of the first hurdles that many clients go through is actually believing that such a migration is actually possible.
So for us in the early stages of engagement with our prospects and clients, we encourage them to take a step wise approach, right, do the analysis, develop a compelling business case, make sure your organization is ready for a transformation.
Other things like doing a four-week proof of concept targeting the most complex and representative workload that you have.
Having clearly defined goals, realistic delivery timeframes from trusted partners, and you know, and so on and so on, like these are all essential tasks, basically, as an exercise in building trust and confidence.
Specifically, typical challenges that we see with inventory collection, data migration and testing, so I’d say for anyone considering a mainframe modernization, make sure your vendor of choice has depth and expertise in each of those areas.
00:16:09 Gary
Now, if we kind of like think about this again more broadly, if you kind of put the vendor claims to one side, from our point of view, you know what matters most to us is that we now have, many customers around the world who’ve been running in production on Heirloom now for several years.
They’ve also provided us with unpaid personal testimonials and in some cases have even written up their own case studies, and that’s something you just do not see very often in any industry.
The point is, you know, we’ve now got a proven track record of delivering successful high value migrations. And so again, I think you know having referenceable clients and make sure your vendor has those referenceable clients for real world mainframe workloads, more than anything else, that’s what matters the most.
00:17:08 Pedro
OK, great. The other part, changing the subject a little bit, let’s talk about the Gen AI. I saw a client recently with questions that I receive are related to Gen AI, can I use Gen AI to convert COBOL to Java and I saw some exercises that do semantic equivalence but not the full application conversion.
But that’s what’s interesting, is that they tried to experiment with the Gen AI and then discovered he cannot do the things, some of the things that I thought it could do. One of the things is converting a full application. How do you see Gen AI impacting the mainframe modernization space? Such as rewriting the applications or creating object-oriented programming languages. How do you see that evolving?
00:18:03 Gary
Yes, I mean, everyone’s talking about generative AI, of course, and rightly so. You know there’s no doubt in my mind that it is a transformative technology, as you know, something on the same scale as the steam engine, the electric light bulb and the Internet.
00:18:22 Gary
However, as you kind of point out, Pedro, I think you know with today’s solutions, it’s really important to understand that generative AI is non-deterministic, in other words, it is fundamentally, by design, it lacks that absolute precision, and basically that means if you repeatedly process the same task, you’ll get different answers pretty much every time.
That’s not a great place to start when, you know, the table stakes requirement of migrating applications is functional equivalence. Not just for the application code, but for everything else that exists around that code. So you know your data, your security, your batch processing, your schedules, right? These are very complex environments and establishing functional equivalence in that type of environment is extremely challenging.
00:19:20 Gary
So generative AI doesn’t actually make a huge amount of sense to utilize it in this context that we’re discussing right now, and especially when you know you have existing technology, compilers and transpilers, that are already the established science and you know, proven to be capable of quickly and accurately transforming millions of lines of code into an object oriented language in minutes.
So I don’t, I don’t think we’re going to see a fundamentally different approach for compiling application code with generative AI for many years yet.
00:20:05 Gary
Now, that said, that’s not to say that generative AI is not impactful in this arena. For us, I think they are kind of three key areas.
The first is to create documentation and insights for monolithic application code. Despite some COBOL proponents claiming that you know COBOL itself is self-documenting and easy to maintain, anyone who’s working on monolithic systems will know that that’s not the case, right? The reality is that these systems and this application code is extremely complex, not well understood, and it makes them very difficult to work with. So I think there’s a real opportunity for generative AI to help establish the code documentation and insights.
The second is the automatic generation of testware and that will speed up the critical testing phases of migrating mainframe applications. You know, I’d say roughly by an order of magnitude, and this matters because today, you know, having now done many, many migration projects, we know that the testing phases typically consume at least 50% of the migration project resources.
The third piece is another accelerator, but this time for the refactoring and modernization of the application post migration. So once you’ve already established something that is functionally equivalent, what do you do with that application going forward?
One example of that, that we’re experimenting with is that Heirloom provides a Java framework that extends the Java platform so it can execute mainframe applications, you know, with that absolute precision that’s required. Now, with generative AI, you can start to replace that framework dependency with something else. Again, just as an example.
Now, of course now you could do that with traditional methods, but the velocity that we’ve seen with generative AI on this kind of refactoring is literally several orders of magnitude faster. So these are the things that we’re already working on.
And you know, to my earlier point about how transformative generative AI will be, the results that we’re getting are pretty stunning.
00:22:44 Pedro
That’s an interesting point of view. Well, I tell you, I have no more questions, but do you have any final words or advice for leaders that are considering modernizing, modernizing their mainframes and their ecosystems? What can you tell clients to do and not do?
00:23:06 Gary
Well, I think if you accept, you know, the main drivers that, you know, you highlighted from your report earlier, right, so essentially cost and but also concerns around agility and a need to move faster, you know, those two things are not necessarily mutually exclusive.
And I think today what we’re seeing is that anybody with the mainframe has recognized that you know these drivers are essentially strategic imperatives and that they need to act on. So for anybody considering this at this point, I would say that you know your competition is already acting on this, so my simplest advice is do something. Like, the cost of doing nothing is increasingly risky to your business.
Just to illustrate that, you know, even IBM today has recognized the need to modernize the mainframe platform and lots of really good work being done around introducing, you know, new tools, modern tools and languages, but also to integrate the mainframe with the cloud.
As I mentioned earlier, because you know the first hurdle to clear is often one of belief, it’s essential to go through a rigorous process of selecting a vendor and a partner that you are confident can deliver.
00:24:34 Gary
If I think about this in terms of project execution, in our experience, delivering these kinds of complex transformational projects requires 3 things.
The first is kind of obvious, but you need great technology that automates and derisks a lot of the heavy lifting and again do that by putting the client in control of the outcomes.
Second, world-class, world-class execution from an aligned team.
And third, clear unwavering executive leadership and support. That’s really essential.
00:25:15 Gary
So there will be challenges along the way, but I’d say that you know, if you can stitch those three things together into a single team of stakeholders from the start of the project through to the completion, you will realize amazingly successful outcomes for your business. It’s not easy, but it is worth it, and again, as I mentioned earlier, our clients are typically seeing their OpEx cut by at least 60 percent.
But perhaps more importantly, you know, with a, with a view towards you know what’s next, what’s needs to happen in the future, they’re now on a platform where they can modernize much, much faster.
00:25:59 Pedro
That’s interesting advice. Thank you for the participation and questions, and that’s all I have now, so Becky, back to you.
00:26:10 Becky
Thank you, Pedro, thank you, Gary and thank you to you for listening to this edition of Heirloom Modcast. We hope you found it informative and insightful.
Now, if you’d like a copy of the ISG report, you can download that from our website. So please go to heirloomcomputing.com/isg2024.
And finally, if you’d like to stay in touch with Heirloom Computing, you can follow us on LinkedIn. So that’s linkedin.com/company/heirloomcomputing. So thank you very much again, and we look forward to you joining another Heirloom Modcast in the future. Thank you.