In two projects led by heimr, one of Brazil’s three largest retail groups renewed the technology foundations of its e-commerce operations without interrupting trading.
- 280 VMs from the group’s sporting goods e-commerce business moved from a physical data centre to IBM Cloud in 2025
- Around 800 VMs across two e-commerce operations moving from VMware to Hyper-V in 2026
- Two delivery cycles completed before the retail peak change freeze
- Second project commissioned at the client’s own request
São Paulo, 2026. Two of the group’s major digital operations modernised their infrastructure in two consecutive moves. The first took the sporting goods e-commerce business out of its physical data centre and moved its workloads to IBM Cloud. The second redefined the virtualisation layer, replacing VMware with Hyper-V within the cloud environment itself.
In partnership with IBM, heimr led both initiatives on one non-negotiable condition: trading could not stop.
In this sector, infrastructure follows the trading calendar. From the second half of the year, every weekend carries strategic weight, and any intervention has to fit the windows the business allows.
“In retail, the deadline is a condition of the business. In the final quarter, the environment goes into a change freeze for Black Friday and Christmas, and every delivery has to be bedded in before then”, says André Frauches, Head of Operations at heimr.
A foundation that no longer matched the business’s ambition
In 2025, the sporting goods e-commerce business ran on around 280 virtual machines hosted in a physical data centre. Ageing servers increased exposure to incidents and limited room to grow.
The IBM Cloud team shaped the initiative to move these workloads to the cloud and brought in heimr to deliver the move, on the strength of a proven track record in critical migrations to IBM Cloud classic infrastructure.
The architecture started from a preliminary design. We reworked that design with the client’s and IBM’s teams and added what the environment required, such as new load balancers and firewalls. We then deployed the new structure on bare metal with VMware and completed the transition on schedule, in agreed windows and with no impact on sales.
“Taking part in the architecture design was decisive. Adjusting the foundation before migrating is what ensures a transition with no surprises”, says Alexandre Lopes, Tech Lead at heimr.
Nearly three times the machines, the same timeline
Also in 2025, changes to VMware licensing under Broadcom meant a significant increase in the cost of cloud virtualisation for the group.
Leaving VMware became the decision, and it is a move that calls for significant investment. The group had to complete the transition within the budget already set, and the project was designed around that limit from the outset.
Hyper-V was the client’s own proposal, taking into account the licensing it already held and compatibility with its systems. With that choice made, three principles guided the project:
- Continuity: no interruption to production
- Performance: capacity to absorb seasonal peaks
- Cost: implementation and migration within budget
“The first time, the challenge was moving past an outdated environment. The second time, it was holding costs without giving up what already worked. More machines, a hard deadline and no margin for downtime”, says André.
Hyper-V: the right fit for the client
The new architecture kept the classic bare metal infrastructure on IBM Cloud. It was the fastest and most efficient way to migrate without changing systems, a requirement common to both projects.
At the virtualisation layer, we compared different technologies before going ahead with the client’s recommendation. Hyper-V required specific tuning parameters to keep pace with retail demand and absorb seasonal peaks.
Technical rigour, working to the retail calendar
Delivery was organised in five stages over around three months:
- Planning and architecture: defining the target environment with IBM. We recommended sizing matched to the real workload, even though it was above the original estimate.
- Build: new environment deployed on IBM Cloud alongside the existing one, with bare metal and Hyper-V ready for production.
- Data replication: preparing workloads for cutover without disrupting day-to-day operations.
- Migration windows: workloads grouped by the priorities of each business area.
- Stabilisation and hypercare: close support until the environment was fully stable.
The most sensitive point of the project was reconciling the schedule with the priorities of each area. We supported the client’s systems team in that coordination, with weekly governance and results presented at the end of each cycle.
“Every priority was legitimate. Our role was to find the fit: which machine moves in which window, without taking away a weekend the distribution centre needs for deliveries”, says Alexandre.
We assembled a team of specialists, with technical leadership and project management dedicated to the migration windows. The client set up a matching team structure, with business areas directly involved. IBM contributed to the architecture and designed the financial model that gave the investment predictability.
A renewed foundation, investment kept in check and trading uninterrupted
Both cycles respected the retail calendar, with no impact on the business. In 2026, the target was to migrate 70% of the environment by 30 September. The entire workload of the sporting goods operation was moved, and the multi-category operation passed 90%, ahead of target and within budget.
“Migrating in retail means understanding our calendar. heimr brought deep command of the technologies and earned the trust of the teams, from systems to the business areas”, says the group’s technology leadership.
Trust earned through delivery
The second project came at the client’s own request, prompted by the way the first had been delivered.
“When we decided to discontinue VMware, the choice of who would lead the project was natural. We already knew heimr’s method and the quality of what it delivers”, says a senior executive at the group.
For André, “critical infrastructure calls for an understanding of the business. When a client comes back and asks for the same team, it means the technology was delivered the way the operation needed.”