Even if IT Isn’t Broken, We Might Still Need to Fix It

In the last six months, the COVID-19 pandemic has forced government agencies at all levels to radically change how they operate. Federal, state, and local agencies were called to update infrastructure practically overnight to support expected requirements – maximum telework, rapid roll-out of new services, and more.

These requirements exposed and amplified legacy technology shortfalls, including challenges related to updating or expanding the use of aging – but still important – applications written in the COBOL programming language. COBOL is still widely deployed on mainframe computers, but has mostly fallen out of favor and suffers from retirements of experienced programmers.

The lessons learned provide a blueprint to accelerate IT modernization efforts, as Federal, state, and private sector experts shared on a recent MeriTalk panel.

“We are good at being reactive and we always get it done when there is a challenge,” Matt Stratton, a transformation specialist at Red Hat explained. “But, the gift COVID has given us is an understanding of the importance of resiliency.”

Stratton says COVID has made it clear that modernization often needs to happen even if everything appears to be working on the surface. This was the case with many COBOL-based systems prior to the pandemic, as an example.

Systems that were stable and working well were put under unprecedented stress. Under these conditions, system overload points became unpredictable. Traffic spiked and leaders recognized the need for rapid modifications. But, due to system age, needed skillsets and knowledge were no longer widely available. Stratton says there are typically gaps in both business logic and technical understanding for these legacy systems.

Krishna Edathil, director of Enterprise Solution Services for the state of Texas, emphasized that moving to more modern agile development methodologies is key, so that IT teams, working collaboratively with business users, can build systems fast and securely.

Edathil shared that for Texas, the first IT challenge with COVID-19 was the switch to remote work over just a few days. Unlike the private sector, just a handful of agencies in Texas had remote work capabilities in place. With support from all levels of state government, 50 agencies were able to move to remote work, some overnight and others over the weekend. “We have the technology,” he said. “We need to develop the art of deploying it faster.”

Kaschit Pandya, deputy CIO of IT Operations at the Internal Revenue Service (IRS), said the agency was not only responsible for this year’s tax filing season (extended in light of the pandemic) and preparing for next year’s filing season, but also for distributing national COVID relief funding and developing a resource for taxpayers to track their stimulus payments.

“The CARES Act passed on March 27, the first set of payments went out on April 10, and the Get My Payment status page was up on April 15 with over 35 million people using it on day one,” he said.

Looking towards the future, Pandya said the team is making that “full-steam approach” the new norm. The goal, Pandya explained, is to improve the pace of modernization. To do so, IT teams need to be able to make changes more quickly – while at the same time managing the associated risks.

You can modernize in increments to accelerate the process and manage upfront risk, advises Kevin Tunks, a technology advisor for state and local government at Red Hat.

Tunks encourages agencies to architect “transition states” and focus on designing solutions to maximize value, scalability, and elasticity. Start with projects where customers and employees will get value from functional changes, he explained. “Make that doable so you can crawl, walk, run – and learn and build momentum.”

“There is a giant application with many moving parts, and you need that collective team to understand it and how it functions,” says Pandya. As agencies work to modernize monolithic legacy systems, he recommends starting by identifying what you can section off. Identify the pieces, what can be de-coupled, and then start to build out the underlying shared microservices. Agencies can then modernize piece by piece.

“How do you eat an elephant?” Pandya asked. “One bite at a time.”

Edathil added that budget cycles and acquisition modernization are also important. He said teams can do a proof of concept quickly, but it might take several years to allocate funding. By then, “a lot of things may have changed.”

The shift from traditional data center infrastructure purchases and license agreements to subscriptions and as-a-service models (IaaS, SaaS, and PaaS) provides new opportunities, if agencies plan with these models in mind. Stratton recommends that teams consider how to design services that can scale up to leverage cloud elasticity and then efficiently scale back down.

In terms of pushing success forward, Pandya emphasized the importance of continuously analyzing the process – the wins, the short cuts they found, and then weave that knowledge into “any and all forward-looking projects…if we have a plan scheduled to take us 24 months, let’s take what we learned and cut that down by 6 or 12 months.”

To learn more – including recommendations for five essential elements of transformation – listen to the full conversation.

Categories

Recent