IG CloudOps had evolved its CloudOps platform through several infrastructure generations – from single VMs to a full Kubernetes setup on AWS. While the platform scaled effectively, cloud costs rose sharply, particularly for compute-heavy workloads.
The team sought a way to regain control of performance, cost, and hardware flexibility without re-architecting their microservices platform.
The Challenge
Even with a well-structured cloud deployment, IG CloudOps faced three persistent issues:
Solution
Working with Scaler Pi, IG CloudOps transitioned from AWS to an on-premise Raspberry Pi 5 cluster. Each device was pre-configured, hardened, and monitored via the Scaler Pi Device Manager.
The new deployment mirrored the previous AWS environment, running Kubernetes on ARM64 across a fleet of Raspberry Pi nodes. Dedicated virtualisation units were included to handle non-containerised workloads. The system integrated with existing monitoring and CI/CD pipelines, ensuring a seamless migration.
Key elements included:
-
High-density Raspberry Pi 5 cluster replacing EC2 ARM instances
-
Hardened OS with built-in telemetry and alerting
-
Prometheus + Grafana observability stack
-
On-prem automation and remote management via Scaler Pi platform
Results
The move delivered a step-change in operational efficiency and cost control.
~67% lower OpEx versus AWS equivalent spend
CapEx recovery within two months of previous cloud billing
Improved density per watt and per £ — higher compute power within the same footprint
Predictable monthly operating costs without reliance on spot markets
Performance remained on par with the previous cloud setup, with improved fault tolerance through a larger node count.
Technical Overview
Scaler Pi handled every stage of deployment — from hardware build and OS hardening to integration with IG CloudOps’ monitoring stack. Each device was shipped ready to join the cluster and automatically register to the management platform, ensuring a consistent, production-grade setup from day one.
Outcome
By moving to Raspberry Pi-based infrastructure, IG CloudOps cut long-term costs while retaining full control over its compute environment. The project proved that a modern, ARM-powered on-prem cluster can deliver enterprise-grade performance without the cloud overhead — setting a new baseline for predictable, high-efficiency edge operations.