August 2026 Summaries

WarpStream introduced Orbit Auto Migration to simplify moving Apache Kafka producers from source clusters to WarpStream without producer restarts, planned maintenance windows, or sustained downtime. Orbit already replicates topic data and consumer offsets, allowing consumers to switch independently, while the new capability uses WarpStream Agents as stateless produce-request proxies so producers can point to WarpStream before a topic is fully cut over. A per-topic state machine moves topics from rejecting direct writes, through proxying writes to the source, into a brief migration phase that blocks and retries producer requests while replication drains, and finally to direct WarpStream operation; failed or slow cutovers automatically revert to proxy mode. The system seeks to reduce migration risks by simulating WarpStream latency during proxying, enforcing WarpStream authentication and authorization early, prioritizing replication for migrating topics, detecting producers still writing directly to the source, and emitting migration audit events. Its design also addresses correctness concerns such as in-flight writes, disconnected Agents, request timeouts, duplicate-versus-loss tradeoffs, and Kafka idempotent producer identities and sequence handling. Available on WarpStream Agent version v826 and later, the feature is presented as a way to perform safer, observable Kafka migrations while preserving acknowledged data and minimizing disruption.

Eternal Ltd., the India-based technology company behind Zomato, Blinkit, Hyperpure, and District, migrated latency-insensitive logging, analytics, and benchmarking workloads from self-hosted and managed Apache Kafka to WarpStream’s BYOC platform. Its previous self-hosted Kafka deployment was limited to one availability zone and minimal replication to control cloud networking costs, while managed Kafka also involved disk, interzone, and lengthy cluster-provisioning costs. Following a two-week proof of concept and two additional weeks to production, Eternal reported 60% savings for its logging infrastructure and 80% savings for benchmarking, while using S3-compatible object storage, avoiding replication and interzone charges, and automatically scaling for traffic ranging from 2.3 GiB/s to 25 GiB/s. WarpStream also reduced benchmarking-cluster creation from 20–30 minutes to seconds and enabled scale-to-zero during idle periods, supporting workloads averaging 52 million messages per minute during backfills. Eternal continues to use Confluent Cloud for latency-sensitive applications, directing workloads between platforms according to cost and performance requirements.