"Legacy is a continuing problem for all of IT."That's the word from Brian Dooley, analyst with Saugatuck Technology, who provided a lay of the land regarding the push and pull between entrenched legacy systems and the demands of digital business.
Of course, this is not news to anyone who's been around IT for any length of time, and understands the issues incurred by legacy systems -- processes and logic buried deep within hardwired systems, likely designed and programmed by people who have long since left the organization.
Keep in mind that legacy is such a relative term. Windows 8.1, for example, can now be considered a legacy system. For definitional purposes, Dooley means mainframes and COBOL-based applications.
There have been many modernization solutions on the market for years that address the problem -- from environments that convert COBOL code into either managed code or re-compiles it into commodity server-friendly languages. Cloud and service-oriented architecture offer abstractions of important pieces of legacy applications to online environments. IBM's own mainframes, the System z and zEnterprise series, support multiple environments, including Linux and Unix, that run in the same box with the same system resources as z/OS.
But has all this modernization been enough? Not likely -- with the pace of digitization, transformation, and disruption picking up, many organizations are still anchored to their legacy investments. Dooley notes that legacy systems still power many critical applications in finance, insurance, government, and healthcare. These systems stubbornly "remain impervious to change," he notes. "Altering these programs is expensive: Results might not be as reliable; the programs are extremely complicated and may be poorly documented; many are heavily patched, creating traps for the programmer, particularly as COBOL skills continue to disappear."
These traps have been around for more than two decades. What's different now than five or ten years ago? Industries heavily reliant on mainframe power -- particularly finance, insurance and healthcare, are at the cusp of severe disruption.
I experienced my own pain as a customer just a couple of years ago -- my auto insurer was handing me outrageous quotes to add my daughter as a driver to our plan. An online search, however, returned quotes from online competitors that were less than half the quoted rate. After several phone calls and threats to move to another carrier, it was revealed to me that our plan was still on the mainframe, which didn't have the capability to spit out more competitive rates. To get better rates, we had to cancel the old plan entirely and set up a new account in the newer systems. Indeed, Dooley notes that the insurance industry "is a huge user of old COBOL programs, yet its customers are now forcing a movement of services to mobile and cloud." Most other customers wouldn't prod with their legacy insurer as much as I did, they would have simply made the switch.
Dooley doesn't offer any easy ways out, because there are none. The work will need to be done on an application-by-application basis, and will take some time. Cloud computing, and API-based development, will offer some ways to step faster into the digital realm, "assuming that APIs can avoid creating legacy problems of their own," he cautions. The solution can only come from on high, with IT and business leaders committing to "smaller systems with fewer dependencies, continuous evaluation of software, and a de-emphasis upon customization."
Organizations will need the will -- and put out the money -- to make the transition. IT leaders and professionals will need to become educators, evangelists and provocateurs, cajoling their businesses into making new investments and doing things in new ways that will keep them ahead of the disruption wave. And perhaps even become part of the disruption themselves.