Client

The Client is a global leader in conversational AI and messaging solutions, enabling brands to deliver human-like conversations across multiple channels. With a mission to make life easier for people and brands everywhere through trusted conversational AI, the Client serves major enterprises worldwide.

Challenge

The legacy architecture faced significant challenges in maintaining consistency and resilience across distributed services. Services within the ecosystem used heterogeneous communication protocols:

  • case-study-arrow-check-black
    WebSocket was used between Service A and B.
  • case-study-arrow-check-black
    A queue-based mechanism connected Service B and C.
  • case-study-arrow-check-black
    Communication from Service C back to Service A had to traverse through intermediate services, leading to inefficient message routing.

This complex and fragmented messaging flow introduced instability, especially during traffic surges or partial network/platform failures. The lack of fault tolerance and a consistent messaging protocol made the system difficult to scale and maintain

Results & Outcome

  • case-study-arrow-check-black
    Reduced Latency & Payload Overhead using Protobuf over JSON.
  • case-study-arrow-check-black
    Improved System Resilience with Kafka handling retries and backpressure.
  • case-study-arrow-check-black
    Increased Scalability through asynchronous, decoupled messaging.
  • case-study-arrow-check-black
    Unified Protocol Stack simplifying service development and deployment.

Highlights

  • 1

    Transition to Protocol Buffers (Protobuf) to serialize data efficiently, reducing message payload sizes and network latency compared to JSON.
  • 2

    Integration of gRPC for internal service-to service communication using Protobuf.
  • 3

    Adoption of Apache Kafka as the primary distributed message broker to enable reliable, scalable, and decoupled message exchange.

Strategy & Solution

To address these limitations, a complete communication overhaul was implemented with the following goals:

Technologies Deployed

Node.js

Backend service runtime

Protobuf & gRPC:

Efficient and typed service communication

Apache Kafka:

Scalable and distributed messaging backbone

Avro:

Schema management for Kafka message streams

Prometheus & Grafana:

Monitoring and observability

Nginx:

Reverse proxy and traffic routing

PM2:

Process manager for Node.js services