Home
TrndX UniFlex™ Platform
All Products Hardware Software
Overview About Us Blogs Careers Team
Contact Us
TrndX Case Study

Our Customer's Home Cloud Journey

How it saved on costs by moving from Azure Cloud & VMware-centric workloads to a homegrown platform.

About Our Customer

Automating the Modern Supply Chain

The customer is an automation technology leader reimagining the global supply chain with an end-to-end, AI-powered robotic and software platform. Its system powers automated fulfillment for enterprise giants across grocery, retail, and wholesale supply chains — orchestrating complex fleets of autonomous robots to process billions of dollars in inventory with maximum accuracy, speed, and density.

Key Highlights

Infrastructure Demands

Rapid development iteration, mixed workload support (VMs, bare-metal, and containers), and high-throughput, low-latency compute.

Engineering Profile

Advanced machine learning, distributed AI workflows, and real-time fleet orchestration.

The Reimagined Cloud

From Split Estate to Unified Control Plane

The customer's AI-powered warehouse automation engineering estate ran split across Microsoft Azure/Google Cloud and a legacy on-premises VMware footprint, with Dev, QA and production environments on separate hardware islands under a rising licensing bill.

TrndX instituted its orchestration platform to create a balance between usage and IT infrastructure needs by building a reimagined cloud inside the customer's own colo-data centre — VMs, Kubernetes and bare metal on one control plane, seamlessly working with the other legacy setup, operated as code. Non-production migrated first, production followed, with no disruption to running workloads.

The result: >50% reduction in cloud ops cost, gaining cloud-like agility, freedom from VMware licenses, and a future-ready architecture.

Key Results

>50%
Lower cloud and virtualization costs
5x
Capacity scaled to 50 nodes on one control plane
Minutes
To provision through code, not weeks
The Starting Point

Paying Twice for the Same Capacity

Split Estate & Double Costs

Non-production workloads were split between Azure, Google Cloud (metered recurring costs) and underutilized VMware (per-core licensing fees), forcing the customer to pay twice for the same capacity.

Fragmented Environments

Dev, Q/A and production environments ran on separate hardware islands, each with its own operational overhead. TrndX was able to design a customized architecture to solve this.

Storage

Existing storage was licensed and not on premises.

The Solution — UniFlex

One Control Plane, Any Workload

The customer's new environment is based on TrndX's UniFlex solution. At its core is the TrndX Orchestration Platform (TOP), a resource-abstraction layer that lets the customer's teams consume compute, unified storage and networking as a service inside their own data centre, independent of the hardware beneath.

The Deployed Architecture

UniFlex Cloud Architecture diagram: two racks (Rack 1, Rack 2), each with MAAS, HA control planes, KVM/Kubernetes/bare-metal/GPU compute and Ceph storage, connected via a 25/100G ToR leaf-spine network fabric and inter-rack uplinks to a shared corporate core network, with isolated out-of-band management and MAAS/Ansible/Terraform/monitoring/CI-CD automation

Two racks, each carrying its own high-availability control plane, KVM/Kubernetes/bare-metal compute, and Ceph storage, connect through a 25/100G leaf-spine fabric with dual inter-rack uplinks back to a shared corporate core network. GPU/AI-accelerated compute and additional JBOF storage live in Rack 2, with reserved capacity held in both racks for future growth. An isolated out-of-band network (BMC/IPMI, Redfish API) handles management traffic separately from the compute and storage data planes, while MAAS, Ansible, Terraform, monitoring/logging and CI/CD pipelines automate provisioning, configuration and observability end to end.

Solution Highlights

01
Multi-rack, TOR-based topology
25G/100G leaf-spine networking with dual core uplinks — capacity grows rack by rack, the way a hyperscaler adds it.
02
Hybrid workload fabric
OpenStack for VMs, Kubernetes for containers, and bare-metal provisioning — all orchestrated through a single control plane.
03
Unified enterprise storage
CEPH delivers block, file, NFS and object services from a single pool — one storage fabric for every environment, built for data-intensive workloads. The customer uses our storage platform not only as part of our solution but also to support its general purpose storage needs.
04
Seamless enterprise integration
Plugged into the customer's existing corporate DNS, DHCP, security stack and monitoring tooling — no change to how IT already runs.
05
Automation-first operations
Ansible and Terraform turn provisioning, patching and configuration into code — repeatable, reviewable, and drift-detected rather than discovered.
06
Built to scale
Modern Intel-based compute, NVMe storage tiers and a high-bandwidth fabric — GPU and accelerator capable, with headroom for the compute and data-intensive workloads already on the customer's roadmap.
A Strategic Partnership Built in Stages

From Capital Project to Repeatable Platform

The customer expanded its engagement in stages, scaling the platform as each phase proved its production reliability — infrastructure shifted from being a capital project to a repeatable, programmable platform their engineering teams can rely on day-to-day and grow over time.

Milestone Scope & Deliverables Business Impact
Phase #1 Stabilizing at Home
  • Cloud migration, installation of the Orchestration Platform, engineering support, troubleshooting and cluster configuration.
  • Ensured speedy migration, operational stability and freed internal engineers from maintenance toil.
Phase #2 Building Private Cloud
  • Deployed OpenStack with Ansible automation and observability.
  • Migrated and stabilized 15 Dev & Q/A environments from Azure to on-premises VMware platform reliance.
Phase #3 Scaling Non-Production Capacity
  • Eliminated VMware reliance & shifted to TrndX Hypervisor virtualization.
  • Scaled capacity 5X to 50+ nodes with full-stack TrndX Orchestration.
  • Delivered a high-performance cloud on a $1.5M hardware stack under Premium Managed Services.
Phase #4 Vertically Scale (Roadmap)
  • Infrastructure efficiencies using the UniFlex solution as the standard template for their production installs.
  • Goal 2027: Deliver a unified platform with compute, storage, K8s and AI/ML Workbench at production scale.

Operating Outcomes

Hours → Minutes
Improvement in provisioning speed
>50% Lower
Virtualization license exposure
Zero
Unplanned production downtime
3 → 1
Environment types unified
Predictable
Multi-year cost fixed at signature
Future-ready
For the next wave of workloads
The Takeaway

Total Cost Control, Fully Utilized

The customer previously struggled with a split estate — paying metered costs in Azure while paying per-core VMware licenses for underutilized on-premises hardware. This fragmented setup created operational silos and high overhead.

No Cloud & Virtualization Lock-In: Migrated off Azure & replaced VMware with TrndX Compute Hypervisor-based virtualization.

Scaled into 5X Nodes: With a clear roadmap toward full production workloads.

Achieved Immediate ROI: By cutting cloud operations costs by over 50%.

Ultimately, the customer gained total cost control, predictability and performance of a fully utilized homegrown cloud — backed by premium management support from TrndX.

Own your cloud. Control your costs.

Schedule a deep-dive workshop with our core infrastructure team.

Get in Touch
www.trndx.ai