When the unexpected happens, will your critical systems be prepared? Here's what to consider while building a disaster recovery plan
When it comes to legacy systems, focus on usability
Just about every large enterprise is at least partly running on outdated legacy systems that are inefficient by modern standards and prevent the organization from taking full advantage of new technologies. Instead of struggling over large-scale replacements and the inevitable pushback from users accustomed to old systems, focus your attention on usability. That advice comes from Peter Yared, CTO of micro app platform Sapho, and former CIO of CBS Interactive. In an interview with The Enterprisers Project, he explains how this can work.
The Enterprisers Project (TEP): Why are so many organizations burdened with so much technical debt?
Peter Yared: Technical debt is there for a reason – it did not make economical sense to upgrade a system. It doesn't make sense to compare it to greenfield operations.
For the daily user of a system who is trained to use it every day, the burden is one of minor inconvenience. For the occasional user of a system, it is extremely inconvenient to use legacy systems for things like purchase orders and time off requests. This is especially true for younger employees. Machine learning can help figure out what is relevant inside of legacy systems and send that service to the occasional users.
TEP: Some IT leaders claim that legacy systems are legacy for a reason – they're robust and reliable. How can organizations find the right balance between adding new features or other innovations and keeping something in place that's been working dependably for a long time?
Yared: I totally agree. If you have a working water heater in your house, why upgrade to a flash water heater, especially if the water heater isn't fully depreciated? When systems work well and are robust there is no reason to rip and replace. However, at some point usability and integration become important enough that it justifies either upgrading the system so that it is more usable, or replacing it entirely.
Generally, legacy systems are tailored exactly to how an enterprise operates. But it should be noted that some enterprises have inefficient business processes that are extremely difficult to change. Sometimes organizations switch an internal legacy system to an external vendor as a way to modernize those business processes, using the new software as a way to force an existing organization to modernize. This strategy can break a lot of glass, but also produce results.
TEP: Speaking of breaking glass, any advice for better managing these changes, which are nearly always painful?
Yared: Rip and replace is very disruptive and expensive when it changes how the daily users of a system use that system, along with the interlocking dependencies between teams based on logic baked into a legacy system. We replaced a lot of legacy systems that were customer facing when I was at CBS and there were always a lot of bumps, as expected. People would get very upset that they had to change how they did their jobs. There were some systems that were so ingrained there was no business case for replacing them. For those systems, you have to look to augment usability.
TEP: What advice would you give CIOs and CTOs about dealing with legacy systems?
Yared: Whether you want to rip and replace or not, understand that it takes a long time. In the interim, try to improve usability at least for the occasional users of a system. For example, there is a very quick win when you can deliver an easy-to-use centralize approvals application process for managers, or an easy way for employees to see KPIs. These are the types of focused applications that you can build on top of existing legacy without too much effort, and they go a long way toward pleasing employees.
We have seen numerous enterprises that still have employees retrieve critical information on mainframe terminals, believe it or not. It's a lot easier to replace those use cases with something quick and fun than it is to replace the mainframe.