Thinklytics

Rackspace Technology · Technology & SaaS · Seattle, WA · 30 weeks

Rackspace Technology System Consolidation case study

A cloud infrastructure provider had 14 business units operating separate data warehouses with no way to share reports across units. Their data infrastructure cost $4.7 million annually. We designed and implemented a federated data mesh that cut costs by consolidating infrastructure but kept each unit’s control intact. This approach enabled cross-unit analytics, which the company couldn’t do before.

Challenge

Over eight years, 14 business units built separate data warehouses with no coordination. This created silos, no shared metrics, no cross-unit reporting, and no way to analyze data across units. Maintaining all 14 warehouses cost $4.7 million annually. The CTO ordered a consolidation, but past efforts stalled because business units resisted change.

Approach

We avoided a centralized warehouse to tackle resistance from business units. Instead, each BU kept control of its data and shared it through a central catalog with standard interfaces. We built the data mesh on Databricks and migrated the BUs in groups of three over 30 weeks.

Outcome

By week 30, all 14 business units were fully integrated into the data mesh. This cut infrastructure costs from $4.7 million to $1.1 million annually. We built cross-unit analytics dashboards that tracked revenue, customer overlap, and operational efficiency for the first time. Switching to a federated ownership model resolved the resistance from business units.

We ended the back-and-forth by handing each team their own data to own.

At first, we tried to pile all the data into one place. But the teams weren’t having it, they didn’t want to lose control. So, we flipped the script. We built a federated mesh setup where each team kept their own data but could easily share it with others. Once they saw they weren’t giving anything up and still got the benefits, they jumped in pretty fast.

How We Tracked Cohorts Together with Business Unit Leaders

Before we kicked off the migration, we grabbed a lead from each business unit. The first three teams dove into the new mesh quickly, which made it a breeze to get the next groups on board. By the time we reached the fourth batch, they were pushing us to move faster.

How We Pulled Reporting Together Across Teams

We put together a dashboard to see how much customer overlap existed between the cloud compute and storage teams. This was something they’d been puzzling over for eight years. Once we laid out the data clearly, even the biggest skeptics had to admit it made sense.

Results

  • $4.7M to $1.1M Annual infrastructure cost
  • 14 Business units on unified mesh
  • 30 wks Full deployment timeline
  • 0 Business unit escalations

We tried to consolidate these warehouses twice before, but it didn’t work because the business units wouldn’t cooperate. With the federated mesh model, that changed. Now every BU is on board, and we’re able to do analytics we never could before.

Chief Technology Officer, Cloud Infrastructure Provider

In the client's terms

The question

What breaks when these systems become one?

The situation

Fourteen business units had built separate warehouses over eight years with no coordination: no shared metrics, no cross-unit reporting, $4.7M a year to maintain. The CTO had ordered consolidation and previous attempts had stalled on business unit resistance.

What we found

The resistance was the design constraint, not an obstacle to overcome. Every prior attempt had proposed a central warehouse, which asked fourteen units to give up control of their own data, and fourteen units had declined. A design that did not require that had never been put to them.

What it produced

All fourteen units were on the mesh by week 30. Infrastructure cost fell from $4.7M a year to $1.1M. Cross-unit analytics on revenue, customer overlap and operational efficiency existed for the first time. Business unit escalations: zero.

The roadmap

Each business unit kept ownership of its data and published it through a central catalogue with standard interfaces, which is a federated mesh rather than a consolidation. Built on Databricks and migrated in groups of three across 30 weeks, so each group learned from the last.

The work behind this result

Related case studies

Frequently asked questions

What is a data mesh and when is it the right choice?

A federated model where each business unit keeps ownership of its data and publishes it through a shared catalogue. At Rackspace Technology it was the right choice because fourteen units had already refused a central warehouse more than once, and the resistance was the design constraint rather than an obstacle to overcome.

Why do data consolidation programmes stall?

Frequently because they ask business units to give up control. Every prior attempt at Rackspace had proposed a central warehouse and fourteen units had declined. A design that did not require that had never been put to them, and once it was, adoption moved quickly with zero escalations.

What does a federated data mesh save?

Rackspace took annual infrastructure cost from $4.7M to $1.1M across fourteen business units in 30 weeks, and gained cross-unit analytics on revenue, customer overlap and operational efficiency that had not existed in eight years.