Bangalore Apache Kafka®, Data and Streams Group

IMPORTANT:
Please make sure to fill out this form for security purposes to enter the venue

***
Agenda:

• 11:00 - 11:10: Welcome
• 11:10 - 11:35: Uttam Agarwal, Engineering Leader, Nutanix & Bhupesh Soni, MTS 3, Nutanix
• 11:35 - 12:15: Parth Agarwal, Senior Software Engineer, Confluent
• 12:15 - 12:30: Break
• 12:30 - 13:10: Avinash Upadhyaya K R, Senior Consultant, Platformatory
• 13:10 - 14:10: Lunch

***
💡 Speaker:
Uttam Agarwal, Engineering Leader, Nutanix & Bhupesh Soni, MTS 3, Nutanix

Talk:
Kafka Resequencer – Reordering the Chaos

Abstract:
In the world of distributed systems, events rarely arrive in perfect order. Network delays, retries, and system hiccups can cause out-of-order events, leading to inaccurate downstream processing.
To address this, we built a Kafka-based resequencing mechanism using the Kafka Streams Processor API. This robust solution:

• Uses RocksDB as a persistent state store
• Leverages Kafka Streams' built-in scheduler for time-based buffering
• Applies versioning logic to reorder events correctly
• Ensures eventual consistency with low latency
• Supports high-throughput and seamless scalability

Whether you're building real-time analytics, transaction pipelines, or event-driven applications, this architecture provides a resilient and efficient path to handle out-of-order data.

-----
💡 Speaker:
Parth Agarwal, Senior Software Engineer, Confluent

Talk:
Metrics pipeline using Kafka and Flink

Abstract:
Confluent's Observability Pipeline for handling metrics leverages Apache Kafka to process telemetry and operational metrics at massive scale. This talk offers a behind-the-scenes look at how Kafka powers real-time observability and analytics for Confluent Cloud, enabling features like monitoring, billing, alerting, and data science across distributed environments. The session will cover architecture patterns, consumption via Druid and MetricsAPI, advanced pre-aggregation and streaming via Flink, practical wins and challenges at scale. Pre-requisites: Basic understanding of Kafka and observability concepts, like metrics.

-----
💡 Speaker:
Avinash Upadhyaya K R, Senior Consultant, Platformatory

Talk:
Unified access layer for Kafka resources

Abstract:
As the adoption of Kafka increases in an organization, a common pattern is the adoption of Platform Engineering with a central team managing Kafka clusters. However, in order to truly decouple the developers from managing Kafka clusters, a common access layer for the Kafka clusters might be necessary to solve for the following concerns:

• The ability to route a client to a specific cluster (bootstrap.servers) dynamically, without having these hardcoded in the client
• The ability to route a client to a specific topic, based on a higher order messaging or routing primitive, without being joined in the hip with a set of topics
• Credential Mediation
• Ability to provide best practices for configuration

The talk intends to propose common solutions that would help overcome these problems and in the true spirit of a platform, provide an effective amount of self service. We will demonstrate one such approach using an open source client-side SDK.