One operating platform.Deployment and observability, built together.
From weekly releases to on demand.
A new Kubernetes platform and 32 migrated services, with deployment control and monitoring built into the delivery.
An anonymized client engagement.
- Engagement
- 12 weeks
- Delivery team
- 2 platform engineers
with an Inpro technical lead
Move the services without disrupting customers.
The B2B software platform was releasing weekly. The work covered standing up Kubernetes and moving 32 application services onto the new platform.
The migration had to account for the running product as well as the new environment. Two platform engineers and an Inpro lead delivered the work over 12 weeks.
No customer-facing incidents were recorded during the migration.
A platform the team could deploy and operate.
The new platform was built in the client’s operating environment while the services moved across, so releases never depended on the old path.
Lay the foundation
AWS and Kubernetes give every service one environment to run in, defined before the first service moves.
Result One environment for all services
Move the services
The services move onto Kubernetes, and Argo CD keeps each deployment in line with the state defined in Git.
Result 32 services, no customer-facing incidents
Release on demand
Prometheus shows how the platform is running, and releases go out when they are ready instead of in a weekly window.
Result Deployment P95 under eight minutes
A clear role for every part of the stack.
Cloud infrastructure, workload management, deployment control and metrics, delivered together. These are the tools used in the engagement.
Prometheus
Metrics & observability. Metrics collection to help the team understand how the platform is operating.
Argo CD
Deployment control. A GitOps delivery path that connects the desired application state to Kubernetes.
Kubernetes
Workload orchestration. The environment for running and managing the 32 migrated services.
AWS
Cloud foundation. The infrastructure supporting the new platform and its application workloads.
ReleasesBefore: WeeklyAfter: On demand
Releases no longer had to wait for the weekly window.
The platform moved to on-demand deployment, with deployment P95 under eight minutes, all 32 services migrated and no customer-facing incidents during the migration.
Results from the anonymized engagement described here.
How to read these results
P95 is the 95th percentile: 95% of deployments completed in less than eight minutes. The incident count covers the 32-service migration.
Results are from this anonymized engagement.
What is holding back your next release?
Bring your platform, the services you need to move, and the constraints around delivery. We’ll work out the next step together.