Field Notes from KubeCon #1: Flipkart's Service Mesh Adoption Journey

Ishan, our Facets team member, arrived early at the bustling KubeCon venue.

After navigating through the crowd and grabbing a quick coffee, he headed straight to the lecture hall for the highly anticipated keynote by Flipkart's tech wizards.
‍

Field Notes on Session: And the Service Is Now Part of the Mesh! A Case Study of Service Mesh Adoption at Flipkart

Presented by: I V Prasad Reddy, Software Architect & Purushotham Malipedda, Platform Engineering, Flipkart

Key Takeaways:

Service Mesh Adoption at Flipkart:
Istio-based ServiceMesh offers a range of features. The scope of these features depends on the level of adoption—whether it's a single service, all services in a call graph, or all services across the organization. Flipkart, with its massive scale of ~6K services and 15 million RPS, faced unique challenges that required innovative solutions.

Challenges and Solutions:
Traditional onboarding methods failed Flipkart. They developed a ZERO touch, operator-powered, phased onboarding model. This model accelerated their public cloud bursting and disaster recovery solutions. Features like service graph and securing east-west traffic were unlocked.

Implementation Details:

• Kubernetes Clusters: They matched the number of clusters to the number of regions to reduce fragmentation, using Cluster API (CAPI) and Argo.
• Service Mesh with Istio: Initial challenges with Istio's sidecar deployment led to using a node proxy.
• Cloud Transition: Flipkart transitioned from a private cloud to leveraging the public cloud.

Impact:
With services onboarded to the mesh, Flipkart can utilize mesh features with minimal and incremental changes. This increased overall usage. The talk covered their journey, the challenges faced, and how the operator model facilitated adoption.

Additional Highlights:

Cell-Based Kubernetes Architecture:

• Microservice vs. Pod vs. Cell:
  • Microservice: A single-function service.
  • Cell: A group of microservices delivering a business capability.
• Scope & Autonomy:
  • Microservice: Independent single function.
  • Cell: Grouping of microservices for broader business capabilities.

Implementation Example:
Saiyam Pathak from Loft Labs gave a hands-on example of cell-based architecture, showcasing its practical application.

Ishan's Insights:

Ishan noted how Flipkart's approach to service mesh can revolutionize large-scale e-commerce platforms. The operator model's success in handling complex deployments with minimal friction was particularly impressive. He predicts that as more companies adopt similar strategies, the industry will see smoother transitions to cloud-native architectures. This session was a goldmine of practical knowledge and forward-thinking strategies.

Conclusion:

The keynote highlighted Flipkart's innovative approach to Service Mesh adoption. Ishan walked away inspired, with a clear understanding of the potential and future of service mesh in large-scale operations.

‍