Skip to content
CloudMagicTalk to us

Retail & E-Commerce

Off five hand-managed VMs and onto containers

A high-volume CRM and order management provider

A high-volume CRM platform migrated out of two data centres, then run as managed operations

900+
data processing tasks moved off hand-managed VMs
Industry
E-commerce / CRM and order management
Scale
900+ data tasks across 5 independent VMs
Engagement
Full migration + hybrid automation + managed operations
Architecture shown
Google Cloud reference design

The challenge

  • 900+ scheduled data tasks ran across five independent virtual machines with no orchestration
  • Infrastructure sat across two separate hosting providers, one of them a legacy data centre
  • Resource allocation for each workload was decided and adjusted by hand
  • Every capacity change was a manual operation, so nothing scaled on demand

What it had to do

  • Consolidate the workloads onto one orchestrated platform
  • Migrate the databases without interrupting a live commerce platform
  • Replace hand-tuned resource allocation with scheduling
  • Keep the platform running afterwards, not just deliver the move

What we built

We consolidated the task fleet onto an orchestrated container platform so scheduling and resource allocation stopped being a manual decision, migrated the databases off the legacy estate, and codified the whole environment so it is reproducible rather than hand-built. We then stayed on to run it.

Orchestrated task fleet

900+ jobs scheduled on a container platform instead of pinned to five VMs.

Database migration

Production data moved off the legacy estate without interrupting commerce traffic.

Everything as code

The environment is reproducible from source rather than hand-configured.

Managed operations

Monitoring, alerting and logging handed over as a running service, not a document.

Reference architecture

Sources

  • Legacy VMs
  • FTP & web tiers

Orchestrate

  • GKE
  • Cloud Scheduler
  • Workflows

Data

  • Cloud SQL
  • Cloud Storage

Operate

  • Terraform
  • Cloud Monitoring
  • Cloud Logging

Results

  • 80 to 90% less manual effort spent allocating resources to individual workloads
  • 900+ data processing tasks consolidated onto one orchestrated platform
  • Databases migrated off the legacy data centre without interrupting the live platform
  • Infrastructure defined as code, so environments are reproducible instead of hand-built
  • Monitoring, alerting and logging in place, with operations run as an ongoing service

For engineering

Capacity is a scheduling concern, not a person adjusting a VM.

For operations

One platform to watch instead of five machines across two providers.

For the business

Peak order volume no longer needs someone standing by.

  • GKE
  • Cloud Scheduler
  • Workflows
  • Cloud SQL
  • Cloud Storage
  • Terraform
  • Cloud Monitoring

Facing something similar?

Tell us where you are now. A senior engineer replies, usually within one business day.

Talk to a customer engineer