Q&A: Mainframe Migration Up Close
Hardware migrations can be tricky. We discuss the keys to the success of Kansas City Southern Railway's migration from mainframe to Linux with Carl Harrison, the company's chief information officer.
Enterprise Systems: Tell us about your company and what drove your mainframe migration.
Carl Harrison: To give you a bit of background, Kansas City Southern Railway Company (KCS) provides a key rail route between the industrial heartlands of the U.S. and Mexico, where our customers benefit from seamless transportation across Laredo / Nuevo Laredo, the largest cross-border gateway between the two countries.
We have around 2.6 million truckloads that originate or terminate in our cross-border target market, with approximately 50 percent of loads moving to and from Mexico. We also operate the rail bridge on both sides of the border.
Just like any competitive business, we're dedicated to providing our customers with exceptional service at the best price. A great way to do this is to reduce our own operating costs, so we can pass those savings onto our customers, a win-win situation.
When my CFO approached the IT department for recommendations for potential cost savings, we realized it was a good opportunity to propose a new approach that would deliver savings both immediately and into the future. Because of this, I looked at the most significant part of the budget -- the operating costs of the mainframe -- and took my decision from there.
What was migrated? (How many applications, what kind of apps, etc.)
The entire IBM mainframe estate was migrated, including the following five core applications:
- Our Management Control System controls all aspects of train movements, rail car scheduling, crew work orders, and waybilling
- The Revenue Management System manages freight revenue and pricing
- A Locomotive Management System controls the movement, train assignments, and repairs of freight locomotives
- Car Accounting is primarily a batch processing system for managing payable and receivables from rail car sharing
- Crew Call is the Enterprise Workforce Management system
Why did you move off of the IBM mainframe?
The IBM mainframe really wasn't old, seeing that it has been bought just seven years previously, but its maintenance and software costs were running at about $3.9 million a year. In addition, five of the company's most mission-critical applications were running on the mainframe and they consumed about 1450 MIPS [millions of instructions per second]. When I realized that reducing mainframe MIPS would bring substantial cost savings to our business, I decided to weigh my options.
We did look at rewriting our legacy applications, but they were all relatively new and already geared for the future. There was just no need to reengineer or change the applications at all. The only viable solution was to "lift and shift" the applications from the mainframe to be hosted on a Red Hat Linux environment.
Did you discuss this move with any outside sources?
I actually worked with analysts from Gartner and Forrester who convinced me that this was the lowest risk way to achieve the kind of cost savings that we were looking for. When I saw that the business case demonstrated an ROI within two years, and an ongoing cost savings of $3.5 million a year -- an unbelievable 90 percent of the cost of maintaining the mainframe -- it was a no-brainer.
What was the scope of the project?
Big, to say the least. The applications involved around 8.5 million lines of source code and 1,850 production DB2 tables, which are used by over 600 people daily, and many others were using reports produced by the applications.
What tools did you select?
We decided to work with Micro Focus' Studio Enterprise Edition and Server Enterprise Edition based on the value we'd be getting in the long run, and the smooth transition that the Micro Focus team said was possible.
What was the migration process like?
Because all applications, aside from the Enterprise Workforce Management system, were developed in-house, we decided to split the migration project into three phases.
Phase one involved moving the four in-house developed applications, including:
- Management Control System
- Revenue Management System
- Locomotive Management System
- Car Accounting
Phase two included moving the final Crew Call application, along with the DB2 development and test environments.
The third and final phase of this initiative will enable us to achieve our goal in turning off the mainframe and replacing the $3.9 million annual maintenance cost with a far more acceptable $400K. This has not been completed just yet, but we expect to be complete before the fourth quarter of 2012.
How long did the first phase of this process take?
Believe it or not, it took us only eight hours to move our four major applications. Not by choice, it had to be done. Our systems across the U.S. and Mexican customs, along with other railroad companies and the transportation authorities gave us just eight hours to complete the cut over. We did it with the help of Micro Focus.
We knew it was the right strategy and the right technology, but up until the actual cut-over, I still had some concerns -- completely without foundation as it turns out.
What's the outcome you've seen thus far?
With phase one complete, over 67 percent of the mainframe MIPS have been successfully migrated to the Red Hat Linux environment, we've cut our maintenance costs by one-third, and we've already seen savings of $1 million just this year alone.
Though the bulk of the savings will be seen when we shut down the mainframe, just moving the four in-house apps to the Red Hat Linux environment, helped us see immediate savings (that's where the $1 million dollars in savings comes into play).
I also want to point out that we've experienced a number of unforeseen benefits as well. Today, our data center has a smaller carbon footprint. This is important as we have a strong commitment to corporate social responsibility. This migration presented us with the perfect opportunity to look at our data retention requirement which lead to an overall data cleanse.
We didn't consider this part of the initial project or cover it in the business case, but we've achieved cost savings here and also within disaster recovery, as there are more hosting partners for distributed systems than IBM mainframes. More competition means we get a better rate.
Any closing thoughts? What are your next steps?
From conversations I've had with my peers, I can see that the concept of mainframe migration is becoming pretty popular among transportation companies these days. I'd like to say that we're something of a trailblazer in this area. In enabling us to move to Linux, the technology and expertise provided by Micro Focus has put us a step closer to the cloud.