Niharika Sakuru
Software Engineer at Bloomberg LP
New York City, New York, United States
Actions
Niharika Sakuru is a Software Engineer at Bloomberg who is currently adulting in Jersey City. She is part of the Compliance Engineering org, working in the team that owns stream processing applications that surveil real-time communication. At the moment, she is working on enhancing the performance of streaming applications. Niharika is a member of the company’s Distributed Systems Guild and Bloomberg’s Women in Technology (BWIT) community. Outside of work, Niharika loves dancing, yoga, traveling, animals, and chocolates.
Links
Area of Expertise
Topics
Storm to Flink: Migrating Latency Sensitive Real-Time Streaming Applications with Zero Downtime
Migrating latency-sensitive real-time streaming pipelines that process millions of requests a day with zero downtime is challenging. Additional system properties such as no data loss, high availability, resilience, etc., make the migration process even more challenging. At Bloomberg, we have streaming pipelines (Compliance Pipelines) to surveil communication traffic in real-time, implemented using Apache Storm. In this talk, we will explore why we decided to move our Compliance Pipelines away from Storm, and why Apache Flink was chosen for this task. We will then go into depth about how we redesigned the application to leverage the framework's strengths. We will also discuss issues we encountered during the migration and how we overcame them. We will look at different metrics that illustrate the performance improvements of the updated application. Finally, we will wrap up with a discussion related to testing, productionizing the application, and going live.
Keep Calm and Flink On: Disaster-Proofing Apache Flink
As real-time data processing becomes crucial for modern enterprises, building resilient and fault-tolerant Apache Flink applications is more important than ever. This presentation explores the foundational and advanced strategies for ensuring high availability, reliability, and disaster recovery in Flink environments.
We begin by demystifying Flink's checkpointing mechanism, its role in fast failure recovery, and common pitfalls to avoid. Beyond checkpointing, we delve into high-availability architectures, from active-passive to active-active setups, and their impact on system robustness. We also discuss the ongoing efforts of the Flink community towards blue-green deployment as a strategy to ensure seamless application updates with zero downtime. Finally, we present Bloomberg's solution for Kubernetes multi-cluster failover using Karmada.
Karmada is an open source project designed to simplify the deployment and management of cloud-native applications across multiple Kubernetes clusters. While it already provides mechanisms for workload failover, it has traditionally lacked support for preserving application state during disruptions, a critical requirement for stateful data processing workloads like Apache Flink.
To address this, Bloomberg’s Streaming Analytics Engineering team collaborated with the Karmada community to extend its capabilities for stateful recovery. This enhancement allows Karmada to not only failover workloads but also resume them with their state intact, making it an even more powerful solution for improving application resiliency across Kubernetes clusters.
By the end of this talk, attendees will gain actionable insights to build resilient, fault-tolerant Flink applications that can withstand real-world disruptions.
What product/technology does the talk focus on?
Apache Flink
Flink Kubernetes Operator
Karmada
Keep Calm and Flink On: Disaster-Proofing Apache Flink
As real-time data processing becomes crucial for modern enterprises, building resilient and fault-tolerant Apache Flink applications is crucial. This presentation explores the foundational and advanced strategies for ensuring high availability, reliability, and disaster recovery in Flink environments.
We begin by demystifying Flink's checkpointing mechanism, its role in fast failure recovery, and common pitfalls to avoid. Beyond checkpointing, we delve into high-availability architectures, from active-passive to active-active setups, and their impact on system robustness. Finally, we present Bloomberg's solution for Kubernetes multi-cluster failover using Karmada, which not only allows it to failover workloads but also resume them with their state intact – further improving application resiliency.
By the end of this talk, attendees will gain actionable insights to build resilient, fault-tolerant Flink applications that can withstand real-world disruptions.
Flink Forward 2025 Barcelona Sessionize Event
Community Over Code NA 2025 Sessionize Event
Flink Forward Seattle 2023 Sessionize Event
Niharika Sakuru
Software Engineer at Bloomberg LP
New York City, New York, United States
Links
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top