Rootconf 2025 Annual Conference CfP

Our Realtime Communication Framework powers millions of concurrent connections across multiple applications, handling over 1 B+ messages per day and scaling to 5 million connections at peak. This talk delves into how we built and scaled this system using Server-Sent Events (SSE) technology, the architecture decisions that enabled this scale, and the challenges we encountered.

The session covers our journey from initial implementation to current scale, discussing our horizontally scalable dispatcher + connection node architecture, which scales on-demand to the number of messages/active connections. The dispatcher and connection machines are highly optimised and handle 4000 messages/sec (1k/core) and 50k connections per node (~10- 15k connections/core) . We discuss our approach to handling peak traffic patterns and how we optimised for cost and performance. We will further share insights into our data storage solutions using ScyllaDB, our strategies for handling connection management, and how we achieved reliable message delivery at scale.

We’ll also explore recent architectural improvements aimed at introducing reliable delivery and ways to debug distributed systems at scale. The talk includes real-world examples of how we tackled these challenges and how you can build a similar system at your organisation.

Key Takeaways

• Understanding architectural decisions and tradeoffs when building a real-time communication system that needs to scale to millions of users
• Practical insights into handling high-throughput message delivery, connection management, and storage optimisation in distributed systems

Target Audience

This talk is particularly valuable for:

• Backend engineers working on distributed systems and communication infrastructure
• System architects designing scalable platforms
• Engineering leaders evaluating architecture decisions for real-time communication infrastructure

Bio

Geetish Nayak

Geetish Nayak is a Principal Engineer at ShareChat with 12+ years of experience in platform infrastructure, focusing on real-time communication and database systems. For the past 2 years, he has primarily worked on building platform frameworks for live-streaming.

Before joining ShareChat, Geetish worked in platform teams at Intuit and Lucidworks (Enterprise Search). He also has entrepreneurial experience, having founded a startup focused on data cataloguing.

Shubham Dhal

Shubham Dhal is a staff software engineer at Sharechat, working on recommender systems infrastructure. Earlier, he was part of the Platform team, focusing on stream processing and real-time communication. His interests lie in real-time data, and he has a hidden agenda to move everything he can get his hands on to streaming…

Before Sharecha,t he was an equity derivatives quant at JP Morgan. Shubham has done his bachelor’s in engineering fromthe Indian Institute of Technology, Kharagpur