We’ve made it to the end of the year—when roadmaps get finalized, dashboards get one last performance tune-up, and everyone quietly hopes Q1 will be a little less chaotic than Q4.
In this edition, we’re zeroing in on what will help you start next year strong: a deep dive into the latest CelerData Cloud BYOC features, a set of performance-and-ops reads to keep close, standout community posts from teams building in the trenches, and a webinar + event worth putting on your calendar.
To close things out with some well‑deserved recognition, we’re celebrating the 2025 StarRocks Award recipients—community members who went above and beyond to share knowledge, ship contributions, and help others succeed!
Watch this hands-on walkthrough of how CelerData Cloud BYOC helps teams run StarRocks for real-time analytics at scale without adding operational overhead. In this session, Kiran Kamreddy (Head of Cloud Product) and Sida Shen (PM) showcase what’s new in the latest release, including smarter group autoscaling, stronger failover controls, deeper observability, improved security and resource isolation for multi-tenant or shared environments, and a live CelerData Cloud tour with Q&A.
Our first session in a Europe-friendly time zone, featuring Fresha —the first company to run StarRocks in production in the UK. The platform has been handling real operational load ever since, and it continues to expand as more real-time use cases move onto the system.
Anton Borisov (Principal Data Architect, Fresha) joins Sida Shen to share what that journey looked like in practice: what worked, what surprised them, and what they’d do the same (or differently) if they were starting today.
They’ll cover:
The problems Fresha needed to solve and the constraints they were operating under
Why StarRocks became the strategic choice to move fast without fragmenting the architecture
What day-to-day production looks like: what scaled, what didn’t, and the trade-offs that mattered
Lessons you can apply when building a production-grade real-time analytics platform
🧪 Performance Tuning, Benchmarking & Disaster Recovery
Getting OLAP query performance right is the difference between “interactive analytics” and “waiting on dashboards.” In this blog, Kaisen Kang (StarRocks TSC Member, Query Engine Team Lead at CelerData) shares practical lessons from nearly a decade building OLAP engines and helping build key parts of StarRocks, including the vectorized query engine, cost-based optimizer, and pipeline execution engine. He breaks down what to focus on first, how to spot the real bottlenecks, and how to build performance intuition that holds up in production.
Performance testing often splits into two camps—speed or cost—but in real life you need both. In this post, Hawk Liu (StarRocks Committer, Software Engineer at CelerData) walks through results from the Coffee-shop Benchmark, an open test suite designed to evaluate databases on compute-heavy joins and aggregations and frequently referenced in comparisons across the modern analytics stack.
Results at a glance:
~2–10× better performance and cost efficiency versus the reference systems (per published comparisons)
Strong linear scalability from 500M → 1B → 5B scale tests
Most queries ~0.5s at 500M and 1B scales; ~1s for most queries even at 5B scale
Even the hardest COUNT DISTINCT workloads (Q10/Q16) finish in ~10s at 5B scale (7.2B rows)
We’ve got a bunch of great write-ups from our users that are definitely worth bookmarking. Huge thanks to everyone who’s shared their hard-earned lessons, configs, and practical pointers for running things in production.
If you’ve ever waited for a dashboard to load long enough to form an emotional bond with the spinner, Josh Cason shares a refreshingly practical post on how they fixed it.
This article dives into how they took loads from 24–48 hours down to ~1 hour by moving off the shared Postgres JSON + CDC chokepoint and into a simpler path—Spark writing Iceberg partitions, then StarRocks serving it (with Temporal/EMR coordinating the jobs). The payoff isn’t just speed: cleaner data modeling, fewer race-condition surprises, and a pipeline the team can ship and run on their own schedule.
Anton Borisov is back with another excellent deep dive, this time on StarRocks Incremental Materialized Views (IVM) and why they matter. It’s a forward-looking look at Phase 1 (merged, not released yet) and how Iceberg snapshot deltas keep refresh proportional to what changed.
Another great read from the Fresha team, by Nicoleta Lazar. It’s a practical walkthrough of orchestrating historical backfills and real-time ingestion—exporting from Snowflake to S3, loading into StarRocks with Pipes, then handing off to Flink for streaming, all coordinated through Airflow. Expect plenty of hands-on detail on schema mapping, YAML-driven automation, and a few real-world operational gotchas.
Manikaran Kathuria shares a detailed write-up on deploying StarRocks on AWS with Terraform + Ansible. It’s a practical, easy-to-follow guide—from provisioning EC2/NLB/IAM with Terraform to the key Ansible tasks for installing, configuring, and running StarRocks cleanly via systemd, plus disk setup and a bit of OS tuning for stability.
📍 Event: The Next Station is...
🗼Tokyo, here we come. If you’re around, join us on
Catch Cheng Sun at the Iceberg Meetup on Jan 21, 2026 (Wed), where he’ll dive into how StarRocks accelerates analytics on Apache Iceberg—including practical optimization patterns (cache + query tuning) to deliver warehouse-like performance on lakehouse data without copying or moving it. 🎟️ RSVP HERE
You’ll also hear talks from Databricks, AWS, and others, covering Iceberg fundamentals, Iceberg V3 updates and migration considerations, and real-world lessons from streaming write implementations.
Join us on Tuesday, January 27, 2026, for an evening of practical talks, real-world takeaways, and good conversations at Open Data Circle — Lakehouse Meetup #2. 🎟️ RSVP HERE
Join us on Tuesday, January 27, 2026, for an evening of practical talks, real-world takeaways, and good conversations at Open Data Circle — Lakehouse Meetup #2. This meetup features deep dives into AI-native directions for lakehouse architecture, StarRocks as a high-performance engine for the AI era, and practical RAG lessons, including an AI Legal Reviewer demo.
Not in Tokyo? No worries—sign up and you’ll get the Zoom link to join online (details will be shared closer to the event via Luma).
The winners are in—and we couldn’t be more excited to celebrate them. Meet the community members who went above and beyond this year: contributors, builders, mentors, and all-around champions who make StarRocks better for everyone. 👉 Check Out the Full List
Better scaling in BYOC, sharper performance and reliability practices, and battle-tested patterns from teams running CelerData and StarRocks in production—hope there’s something here that helps you hit the ground running in the new year.
One last thing: we’ve refreshed the CelerData website, so you can explore the latest content, updates, and resources—be sure to check it out.
Here’s to quieter pagers, faster dashboards, and a holiday season with zero surprise interruptions!
Prefer updates in your feed? Subscribe to our LinkedIn newsletter.