Apache Iceberg Virtual Meetup: Manifest Surgery: Improving Scan Latency at Scale with iceberg-rust

​We welcome you to the second virtual meetup in the new Apache Iceberg Virtual Meetup series.

This month we'll focus on one main presentation:

Manifest Surgery: Improving Scan Latency at Scale with iceberg-rust
w/ Liam Earley

At Zus Health, the platform team found itself needing to optimize a specific (cost & latency-sensitive) query pattern against TB-scale merge-on-read tables receiving a high volume of micro-batch writes. The usual levers (table partitioning & compaction) were reasonably well-tuned already. After evaluating several query engines, we settled on iceberg-rust for cost/simplicity. To further reduce latency/memory overhead, we used iceberg-rust to optimize manifest layouts for this access pattern without impacting other consumers of the table.

​And then we'll open the floor to hear from you, the community, about what you've been working on and invite you to give 5 minute lightning talks after the main presentation has completed.

Join the Community
Continue any conversations via apache-iceberg's slack. We've got a channel dedicated to this under #meetup-virtual.

Want to Present?
We've created a google form to simplify the process of submitting a talk. This is a rolling CFP, and you can submit as many ideas as you want. Please, refrain from submitting AI Slop (yes it is easy, but we want to hear from you).