Three years ago, Red Hat decided it was time to modernize.
The company had only recently been acquired by IBM in a $34 billion deal that closed in July 2019. The industry leader in open source enterprise software was emerging from one major change and preparing to leap headlong into another.
“The key business systems that ran the company were pretty much on-prem,” said Jim Palermo, who was a senior IT senior director in Red Hat’s digital solutions delivery division when 2020 began and a VP of IT at the company by year’s end.
“We were pretty locked into that data center environment, meaning that there wasn't a lot of ability for us to do any kind of transformation in place,” Palermo said in an end of September interview with CIO Dive. Just two weeks later, on Oct. 12, he was named the company’s CIO.
In August, Red Hat shuttered its legacy data centers, according to Palermo, the culmination of a long journey to hybrid cloud.
There are lots of good reasons to modernize — efficiency, optimization, operational speed and savings rank high on the list. There are challenges too — if it were easy, migrations would happen overnight.
Moving dozens of applications onto a new platform and learning new technologies without disrupting the day-to-day flow of business takes patience, planning and persistence, even for a company with the in-house technology assets of Red Hat.
“We had some advantages, but we were not immune to the things any company is going to face when they go through this,” Palermo said. “We obviously have access to our product engineers and business units. That makes it a little bit easier. But we were just one customer of many, and we did not get prioritized over all the rest.”
An app-driven approach
One of the first questions Palermo said he asked himself during the planning phase was, “If we were to start over, what would we be looking for?”
Red Hat needed on-prem systems that could be integrated with public cloud. Low latency between the two was a must. That answered the infrastructure question.
Applications were a heavier lift. It took nearly a year to review the company’s app portfolio and decide the fate of each one, according to Palermo.
“You're going to end up doing a lot of investigation and recon and looking under a billion different rocks because you need a complete inventory of key business systems and how they're dependent on one another,” said Palermo.
More than 30 app teams were deployed in the strategy phase to accomplish three goals:
- Update a configuration management database to include all applications running the legacy environment.
- Review each application to determine its fate: retirement, migration or modernization.
- Develop what Palermo called a “flight plan” for getting each application from point A to point B.
“We definitely took an app-driven approach,” Palermo said. “Most people think it's an infrastructure thing. It's really not. Infrastructure is your foundation. But at the end of the day, you have to work with your app teams to figure out what it's going to take to move your workload from here to there.”
Migration overload
While there were challenges in the planning phases, implementation took a bigger toll on Red Hat’s technologists. Over the course of a year and a half, Palermo and his team managed “waves and waves of migration,” he said, leading to periods of frustration and exhaustion.
There were integration issues at the network level, but the biggest problem was migration overload, according to Palermo.
“As you move things over, you start to find issues,” he said. “We were working through those issues, upgrading different parts of our software stack, and that forced [the application teams] to deploy and redeploy. There was a lot of exhaustion around that.”
Palermo empathized. “If you're an app team, you want to be building things, right? You don't want to be redeploying and upgrading,” he said.
As the doors closed on Red Hat’s legacy data center in August, Palermo could point to roughly 150 retired applications and several other major wins.
“We were able to transform about 20% of virtualized workloads to containerized workloads and about 20% of bare metal workloads to virtualized workloads,” said Palermo. “We have an on-prem environment that literally has hyper latency — the next cage over is Google, the next cage over is Amazon and the next cage over is Microsoft.”
Palermo’s team architected a hybrid cloud infrastructure with distinct availability and security zones, which allows them to make changes in the overall stack without disrupting operations in IT or the rest of the business, Palermo said.
It didn’t hurt that they had an inside track on Red Hat assets and access to in-house digital transformation specialists. Even that didn’t obviate the need for a little grit, some determination and a good bit of ground-level planning.
“We leveraged Red Hat software for a lot of its architecture and Red Hat people, which is an advantage,” Palermo said. “But if you don't have a good sense of your application inventory and a really strong understanding of how your applications are built, it's going to be really tough.”
