Migrating Our Data Center to a Best-in-Breed Colocation Facility

IT is a constantly evolving field. Today’s hot technology will soon be yesterday’s news. The landscape is always changing, and IT professionals have to adapt. When I joined Sentry 17 years ago, some of the top trends and buzzwords were data warehousing and data management. One of our initial objectives was building operational data stores and new business intelligence and reporting capabilities.

That was 2003. By the end of the decade, attentions had shifted to IT infrastructure. And if you look at infrastructure in 2021, it no longer resembles what organizations used in 2010. 

Technology Transformations Can Take Years

As a personal and commercial insurance company based in Wisconsin and serving policyholders nationwide, Sentry strives to balance ease of customer access and functionality with stability and security to protect our customers’ data. As such, we know we need to adequately invest in our IT strategy, and that strategy must be carefully thought out. 

Much of our transformation traces back to the willingness of our executive leadership to invest in Sentry technology and turn it from a liability to a strength. Our CIO at the time built a new IT team and replaced every aspect of our application portfolio, including HR, finance, policy administration, billing, and claims.

We also adopted a strategy of improving data—a strategy that’s still in effect today. Each time we update a system, we also build or update a data store to accompany it. But we don’t stop there. We continue to reap the benefits and build on the transformation from the past 15+ years, which took us from monthly reports to near real-time analytics. 

But as our infrastructure grew to keep pace with our system capabilities, we found ourselves rethinking how we’d architected our data centers. 

Our Data Centers Weren’t Best of Breed

In 2015, we had two traditionally run, fully redundant data centers within geographic proximity of each other. They were self-managed by Sentry, which means we paid for everything, including cooling, electricity, physical hardware, and the actual buildings themselves. These data centers were good enough, but we wanted to do better—by parking our best-of-breed infrastructure and system capabilities in best-of-breed data centers.

One of our goals in migrating data centers was reducing physical access to our primary data center. We wanted to establish adequate separation between our IT assets and our associate population. There were also structural concerns. We’d modified the plumbing, cooling, heating, and power within the data center to protect our hardware and keep things running in the event of an outage, but the data center wasn’t housed in a purpose-built facility. To make everything foolproof, we wanted to move our equipment into a colocation facility that provided optimal environmental conditions and full physical and IT security.

At the same time, we started thinking about moving some of our systems and infrastructure to the public cloud. We wanted to disentangle ourselves from the day-to-day management of physical infrastructure. But we knew one of our first priorities needed to be finding the right vendor to help us migrate our data centers to colocation facilities. 

Partnering with OneNeck to Migrate our Data Center

A clear frontrunner emerged early on. OneNeck stood out from the pack with their combination of expertise, an expansive technology portfolio, and a culture aligned with the way we do business here at Sentry. We already had a good working relationship with OneNeck’s parent company, TDS—a long-time client—so we decided to delve further into their subsidiary to confirm it was also a good match.

OneNeck had several points in its favor. First, the company is local to Wisconsin. We have an office in the Middleton area, and a number of associates in Madison, where OneNeck’s data center is located. If we needed to sit down for a face-to-face meeting or visit the colocation site, it was just a short trip away. 

Additionally, the team at OneNeck struck us as extremely professional. When we toured their Madison data center, they ticked off every item on our checklist in terms of redundancy and physical security. Their facility was spotless, and we got the sense they were not only disciplined but accustomed to working with top-tier organizations.

Finally, their account team was highly responsive in fielding our questions. We threw a lot their way and asked them to address our short-, medium-, and long-term concerns. When you start working on a colocation strategy, you have to plan ahead. You can’t guesstimate and then try a different approach three months down the line.

A Phased Approach

After we signed the contracts, we proceeded with a phased migration. Phase I was planning. We went through our capacity plan with OneNeck and determined the number of racks and amount of floor space we needed. This wasn’t a two-day exercise—we weren’t scribbling vague ideas on the backs of napkins over drinks. It was an extended process of working out what we needed right then, six months later, and five or six years after that, including storage arrays, servers, network connectivity, and redundancy.
 

We costed everything and moved to Phase II, which was installation, followed by testing. We set up the data center at OneNeck’s colocation facility and then tried to break it. Once we were confident all the pieces were working as they should and were adequately secured, we moved to Phase III, our Data Center Migration project, which took a couple of years.

At Sentry, we manage our application portfolio by business unit or subject area. That means we handle our claims processing applications and infrastructure as an entire ecosystem. The same goes for personal lines insurance, commercial insurance, policy administration, billing, and our customer and internal web portals.

We started moving systems to the OneNeck colocation facility, one at a time, in 2017. While our initial goal was to split our systems 50/50 between OneNeck’s Madison data center and the public cloud, by 2019, we’d adjusted this target to an 80/20 split favoring the colocation facility. We know that number will likely change in the future as cloud technology evolves to meet our requirements.

One of the biggest benefits of the public cloud is scalability. If we need more storage and compute, we can spin it up in seconds. There’s also far less to manage on our part. AWS and Microsoft Azure make sure the infrastructure runs as it should. When infrastructure issues arise, they’re often resolved by the cloud providers behind the scenes while our assets remain 100 percent functional. That said, I doubt we’ll entirely move to the cloud. Our business requirements likely will dictate that we maintain some physical infrastructure, be it on-prem or in colocation facilities. 

Improved Security and IT Hygiene

The shining star of this story is the elimination of our on-prem data center. I can sleep at night knowing that it’s one less asset for us to manage, and I no longer need to worry about some of the unique challenges related to managing power and cooling.

Having a physically separate data center has also forced us to improve our IT hygiene habits and streamline our infrastructure management processes. Our infrastructure associates are more disciplined [SC1] than they were before and tend to plan ahead because they can’t just waltz into our colocation facility to make a change. 

A Comfortable and Cost-Effective Partnership

We’ve established a comfortable working relationship with OneNeck. I’m glad we no longer have to handle the facilities management side of the equation. They maintain the air conditioners and backup generators, and they test everything to make sure our servers don’t overheat or lose power in the event of an outage. OneNeck also negotiates with network carriers to ensure redundancy in the event we lose connectivity. They have three or four cross-connects with various ISPs, and we no longer have to worry about our sites losing our link to the backbone of the internet and the rest of the world. 

As a bonus, letting OneNeck do the heavy lifting is far more cost-effective than building, managing, and maintaining physical data centers ourselves—but we still had to do a lot of planning on our end. Once we’d established our IT infrastructure and data center requirements, we handed the reins to OneNeck. They built a colocation solution that was fast, stable, competitively priced, and easier to manage than an on-prem data center. 

Proceeding with Caution and Avoiding Shortcuts

Here at Sentry, we like to proceed with caution when embarking on technology transformations. We take time to build trust and establish comfortable working relationships with our IT vendors. This slow and steady approach may not yield instant results, but it allows us to evaluate our business model and forecast capacity, helping us prevent costly mistakes.

By avoiding shortcuts, we successfully migrated our on-prem data center to a colocation facility without impacting our daily activities, and laid the groundwork for future migrations to the public cloud. We put our faith in OneNeck, and they rewarded us with a data center that gives us a competitive advantage because we can focus our activities on serving our policyholders instead of servicing our infrastructure.