Scaled Kubernetes Continuous Delivery with GitOps

Authors:

Table of Contents

About Ruckus Networks

Ruckus Networks is redefining connectivity by bridging the digital divide and connecting people around the world. Ruckus works with channel partners, OEMs, and strategic partners to deliver ubiquitous connectivity with our access points, switches, and cloud services.

Migrating from Monolith VMs to Kubernetes Microservices

At Ruckus Networks, past monolith application environments (e.g. Dev, QA, Staging, Prod) could take 30 minutes or more to spin up, with DevOps and development teams left waiting.

As a result, horizontal scaling and elasticity of services running in production became a major challenge.

In addition, deployment pipelines were service specific, with little to no reuse of logic or configuration across teams.

A decision was therefore made to migrate to a cloud-native Kubernetes-based microservices architecture, where environments could be provisioned in seconds, and horizontal auto-scaling was embedded natively to the run-time environment.

As a result of this migration, 40+ new Kubernetes microservices would all require new deployment pipelines and onboarding by the DevOps teams.
“Onboarding a new Kubernetes Application could take our DevOps engineers one week,” said Raj Zala, Director of Cloud Engineering at Ruckus Networks.

Build vs. Buy For Continuous Delivery?

To solve this onboarding challenge, Raj and her DevOps team looked at several options:

Manually script each Kubernetes pipeline – ~40 weeks of effort (1 week per microservice)
Build an MVP CD platform – ~52 weeks of effort (2 DevOps engineers for 6 months)
Purchase a commercial Continuous Delivery solution

Onboarding a new Kubernetes application could take our DevOps engineers one week.

Rajeshwari Zala | Director of Cloud Engineering | Ruckus Networks

Leveraging Harness GitOps and Pipelines-as-Code

One of the key selling points of Harness was to give DevOps and developers self-service Continuous Delivery capabilities, meaning they could empower, automate and scale their software delivery process from dev all the way to production without waiting more than a few minutes.

The second advantage was the ability for the DevOps team to build and reuse pipelines across their portfolio of new Kubernetes microservices and engineering teams.

“Our DevOps team now creates pipeline templates in Harness, and developers instantiate these with their own service parameters,” said Raj.
Harness GitOps allows DevOps/developers to manage pipeline, service, environment, Helm, and Kubernetes configuration all in one place as code. This means all pipelines can be version controlled and managed like code in any Git repository.

We can now onboard new Kubernetes microservices same day instead of one week