Managing a Petabyte-Scale Data Lake with Granica

Join us for a deep-dive with the team at Granica as they walk through the architecture they've built to tackle Data Lakes at enterprise scale: a unified maintenance suite covering vacuum, partition expiration, and right-to-erasure across all five major table formats. Expect real deletion semantics by format, common migration footguns, and the safety layer — retention floors, deferred recovery, multi-schedule, S3 Inventory — that makes cleanup safe at petabyte scale.

Tarang and Tom will walk us through the architecture, share the deletion semantics that differ by format, call out the migration footguns teams consistently hit (shared-path tables, deletion vectors silently breaking VACUUM), and explain the safety layer — retention floors, deferred recovery, multi-schedule, and S3 Inventory integration — that makes it possible to run aggressive cleanup at petabyte scale without losing sleep.

What you'll take away

• Deletion semantics across Hive, Delta, Iceberg, and beyond
• Common migration footguns — shared-path tables, deletion vectors breaking VACUUM
• How to make GDPR right-to-erasure auditable at scale
• Safety patterns: retention floors, deferred recovery, S3 Inventory integration

Speakers:

Tarang Vaish — Co-founder & CTO, Granica Tarang built production AI email security at Armorblox (acquired by Cisco), scaled distributed data at Cohesity, and worked on GPU development at AMD — giving him a full-stack view from silicon to SaaS. Granica now runs in production at some of the world's largest SaaS companies, managing 100+ PB of customer data. M.S. Computer Science, Stanford.

Tom Molloy -- Director of Enterprise, Grancia — Tom was one of Snowflake's earliest sales hires and spent nearly a decade growing the platform from pre-beta to enterprise ubiquity. He later led go-to-market at dbt Labs and theom, working with large organizations on data governance, transformation, and security at scale.