Technology leader, founder, engineer, consultant, trusted advisor, speaker. I'm passionate about technology, and love helping businesses get the most out of public cloud.View the profile
About the talk
Mainframe powers much of global commerce and for decades—with its proprietary platform and legendary lock-in—was resistant to effective competition. Even years after most organizations began adopting public cloud, migrating off the mainframe remains too complex for many organizations to undertake.
Google Cloud brings a unique, automated approach to modernization enabling customers to go from mainframe to containers. This session talks through the challenges of modernizing mainframe and the key technical aspects of this solution that make it possible.
Speaker: Travis Webb
Google Cloud Next ’20: OnAir → https://goo.gle/next2020
Subscribe to the GCP Channel → https://goo.gle/GCP
product: Kubernetes Engine; fullname: Travis Webb;
event: Google Cloud Next 2020; re_ty: Publish;
Welcome to Mainframe modernization. Accelerating Legacy transformation with Google Cloud. I'm Travis, Webb Enterprise Solutions, architect at Google, thanks for joining me at Google Cloud. Next on-air, I work with our customers to help them solve their biggest. Most complex it problems and today, I'm going to discuss Mainframe modernization. A new solution offering for Google. This is an introductory talk. So we're going to start out with an overview of the Mainframe ecosystem. All been discussed mainframes place in our
world today and some common ways that our customers have tried often with bad results to migrate off the Mainframe. Next introduced Google's offering for Mainframe modernization and finally, we walk through one customers Journey as they migrate, thousands of as400, Mainframe COBOL program to run natively in Google Cloud. Let's talk about the Mainframe ecosystem. First, let me give you a quick overview of the Mainframe market today. The Mainframe Market is large,
includes the cost of the machines themselves, but also the fees paid for the usage of those machines and licenses for Mainframe specific software. The vast majority of large companies use mainframes for to run critical business workloads, and many of those applications have been built over the course of decades Mainframe usage, extends beyond the very largest companies, and there are many thousands of Mainframe systems still in use. In each of these systems represents an opportunity for Google and the customer to come together on
a strategic transformation of their business. As commodity, workloads are moving to the cloud at rapid Pace Mainframe workloads, have lagged behind. Let's talk about some of the reasons for this. It's important to remember that for the most of the history of computing. The idea of running important business processes on a personal computer would have you laughed out of the room. The PC was a toy. It was only in this current Millennium that doing important, things on individual, PCS or servers was
taken. Seriously. So how is the Mainframe? So resistant to effective competition? First to IBM Mainframe has an unbroken chain of backward. Compatibility going back to the 1960s, this durability and allowed the Mainframe to amass, a large installed base of customers, and a reliable Revenue stream for Mainframe fenders. And while the physical systems from the 1960s are no longer in use today, the early programs written decades ago, survive to this day,
Second, it leaders have a tendency to closely scrutinize the entrance costs. When evaluating technology, purchasing decision, while much less attention is paid to the exit, leaving the main frame platform with expensive and difficult for many businesses. The cost of migrating still isn't economically viable. A proprietary ecosystem adds to the high exit costs because many Mainframe applications rely on proprietary technologies that are not available outside the Mainframe ecosystem, this combination of factors lead current Mainframe
customers with few viable alternatives. Is the Mainframe. Vendor has a large installed base of customers who can't leave? Why should they invest in innovation? Wow, much improvement occurred throughout the twentieth century, the relative advantages of the Mainframe over. Other platforms were routed over time, and today Mainframe customers. Now, find themselves at a technological and cocked disadvantage compared to their peers. You no longer need a Mainframe to run intensive processing tasks, but if you already have one,
you're stuck paying for it. Anyway, the interesting thing here is that these disadvantages are widely recognized and understood but Mainframe customers had few viable options available to assist in migrating their Mainframe. We're close to the cloud. With no new customers or developers are entering the ecosystem Mainframe. Vendors be emphasized investment in innovation. Of course, companies started figuring out how to use commodity servers to run business work clothes instead of mainframes Google of
course was at the Forefront of this movement and has pioneered, many T data center infrastructure Innovations. We serve customers all over the world with technology that we built into Google cloud and it's at a scale that simply would never have been possible with mainframes. Meanwhile, public cloud is democratizing it Innovations at an incredible pace and its explosive growth. Comes from companies looking to leverage modern technologies that their legacy vendors aren't providing.
It's no secret to anyone that Legacy it is expensive and Mainframe popping tops the list but lots of things are expensive and high operating costs might be tolerable. If you could also got Innovation and Agility in return but with Mainframe, you really end up with the opposite. 8, static ecosystem, that gets more and more expensive and makes it harder and harder to innovate. Customer, expectations are increasing. The pace of innovation is increasing yet many Mainframe customers. See a chronic shortage of qualified staff. Who understands these
Technologies putting at risk Decades of investment in the platform. Again, these issues are widely recognized widely understood most acutely by Mainframe customers themselves. And many are saying we want out companies. Have tried, many different approaches, over the years, to migrate off the Mainframe platform and often never quite achieved escape velocity in the end up getting pulled back in. There are three main ways customers attempt to modernize these Legacy systems on their own. You can undertake
a large manual, rewrite of all your Mainframe software into the language of the week, you can use an emulator to run your entire Mainframe environment on normal computers or you can modernize in place keeping the Mainframe but updating software. We're going over these approaches because they are all antipatterns and we want to help customers avoid these pitfalls in the future. I talked to customers in the middle of a manual rewrite who have been modernizing since Y2K it's expensive whiskey and it takes forever.
And if you work with developers, you know, it's difficult for them to resist the temptation to improve things along the way. This, of course, sets off a chain reaction of yet, more software and business process changes. Further prolonging the effort mean, while the technology around, you is changing all the time. The idea of having a clean slate on a fresh start sounds great, but unfortunately experience shows that almost never works. What about emulate? The Mainframe environment while you
can't achieve True. Cloud-scale Performance with a single Mainframe pound-for-pound, it's actually really good. Hardware. Most of the problems lie in the ecosystem layer and in the cost structure to bike emulating your Mainframe software to run on different Hardware. You really throwing away. Probably the only good thing about the system and keeping all the stuff that's actually holding back. You would curb many of the same risks as in other, migration approaches, But realize few of the benefits. What about Institue modernization? We also see
companies try to modernize in place, they conclude that their own software is the primary problem and needs to be updated. At the same time, they attempt to bring a pseudo Cloud capabilities to the Mainframe This doesn't solve any of the problems of cost or Talent scarcity, and it still involves a large manual effort to update existing software. It's a very expensive way to prolong the inevitable. Google. We had a different perspective. We talked to her customers, we feel their pain and they've been looking to their Cloud partners
for way to migrate these Mainframe workloads to the cloud, we've built a solution called G4. It combines Google's considerable engineering capabilities with Decades of Mainframe modernization experience. We've developed expertise and Technology internally as well as through key Acquisitions. What is g for the G4 platform is a set of tools developed and proven over the last 25 years that transforms your legacy Mainframe applications to run natively on Google Cloud by converting Legacy
code to job. Based on Open Standards with G4, we successfully taken customers from COBOL two containers in a matter of weeks. What was previously too expensive to risky? Too difficult is now possible at Google scale and we believe that this is a unique offering not available elsewhere in the market. Let's talk about G4 for a minute and go over what it does and how it works. G4 consists of three major components, the analyzer, the converter, and the dashboard,
the G4, analyzer purses, your main frame source code to understand your business logic and program dependencies and relationships. The G4 converter applies, conversion, rule to your Mainframe code and transforms it into modern Cloud native Java. The converter supports dozens of Mainframe languages, such as COBOL, Pia one, RPG, JCL, assembly, and many others and can also help modernize database business logic. I M S & V Sam. The G4 dashboard provides performance and complexity metrics
about your software that assists in the testing and tuning of the applications. At a high-level, our migration process follows the tried-and-true methodology that we use for migrating other mission-critical workloads to Google Cloud. The insights, from the G4, analyzer Drive the discovery and assessment phases, which feed into the G4 converter, which modernizes Legacy Mainframe code. Proper validation, and verification, and test ensure that the business logic operates as expected and reduces risk. The chief or analyzer
purses, and decomposes your sources into clusters, which informs the optimal migration strategy G4. Then converts the Mainframe sources cluster by cluster and produces Target code in Java that is optimized to run in the cloud. This collector, based strategy allows the customer to migrate their applications in waves and paralyzed in migration effort using a migration Factory approach. And I should know to hear the code produced by the G4 converter is free and clear, no license, No flocking, No
Nonsense. Let me summarize some of the important than if it's that customers can realize with G4, you can reduce costs and risks, while increasing agility and Innovation costs. Decreased when you escaped the capital-intensive Mainframe, refresh cycle eliminating, its charges in end, Reliance on undifferentiated maintenance Toya, just to keep the Mainframe running. Business risk is reduced by ending Reliance on a closed, ecosystem, unsupported software and a chronic Talent
shortage G4 increases operational, efficiency, agility and Innovation like extending the life of mature software assets, running them on the cloud and leveraging existing investments in your software. By gaining access to Modern Cloud Technologies, you can further optimize adapt and evolve your software in ways that were not previously feasible. With some of these benefits in mind. I like to share with you the Journey of one of our Mainframe modernization customers. We're working with a large Credit Data.
Aggregation company. You can think of them as a credit bureau, who is anticipating a 5x increase in customer queries due to recent changes in legislation? Most of their mission-critical applications are written in cocoa and running on both a Mainframe and a S400 style system. They're also making use of other Mainframe Technologies like kicks JCL, easy trees and DB to increase in demand. The customers looking to migrate to a modern scalable platform and to modernize their applications to support future enhancements. They
had thousands of as400 Mainframe programs in 12 months, to make it happen. What kinds of workloads did we migrate in Mainframe Land? There are two major categories, workloads online and batch and they had different architectural requirements, online programs, process requests and responses in real time and record those transactions in a database as they happen. So, if I want to transfer $100 into my bank account now, and that would be an online transaction. This is
also known as oltp or online transaction processing. Batch programs process book them off into a great records generated throughout the day and perform analyses as an example of a batch process. Might aggravate all trades that occur in a given trait and determine A bank's risk. Exposure programs are scheduled using JCL job control language. Add be modernized, JCL bad programs to do the expense ability of G4, we can create custom conversion rules in this case, we
modernized the customers JCL to run Cloud composer. A managed workflow orchestration service in Google cloud based on pechi airflow, these work clothes trigger batch process is running on compute engine container. Optimize the VM inside managed instance, group has our Java executables running the openjdk runtime environment. online clothes, we deploy, the converted Mainframe applications to compute engine inside a managed instance group Because these workloads are intended
to be accessed by partners and customers over the public internet. We deployed a load balancer to direct traffic to the VM and we migrated the existing db2 database to db2 luw in compute engine. Depending on your access requirements, you might instead deploy a VPN or an entry connect to your data center for secure communication, within your corporate Network, and depending on your transaction processing requirements, we could deploy db2 luw directly to Google cloud or we can replant for me, a relational database to postgres
and deploy. Our Cloud SQL, managed service. Okay, we covered it pretty wide range of topics in this talk from Computing history, to Cloud, economics to Mainframe deployment architectures. So, here are a few key points. I'd like to highlight before you go. G4 is a true end-to-end Mainframe migration solution from Google. It enables, our customers to reduce costs and risks, while increasing agility and Innovation, concepts are in high demand by our customers. Due to the potential, this has to
transform their business. And importantly, Google is the only cloud provider with an integrated into end Mainframe modernization solution. Thank you for joining me today at Google Cloud next on air. I look forward to working with you on your Mainframe modernization, Journey.
Buy this talk
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.