Cover Image for Apache Gravitino 1.3 Deep Dive: Govern at the Source
Cover Image for Apache Gravitino 1.3 Deep Dive: Govern at the Source
Avatar for Data For AI Events
Presented by
Data For AI Events
Hosted By

Apache Gravitino 1.3 Deep Dive: Govern at the Source

Virtual
Registration
Past Event
Welcome! To join the event, please register below.
About Event

Your data estate is already plural, and it got that way without anyone deciding it should. One team standardized on AWS, another built on GCP. An acquisition showed up with its own lakehouse and its own catalog. Glue came free with the account and quietly filled up with tables. Nobody planned it. It accumulated.

The usual fix is to consolidate: pick one catalog, copy everything into it, migrate the estate. It rarely goes well. The copy becomes a second source of truth that drifts from the first. The data owner's access controls get re-implemented somewhere else, by someone else. Your audit trail splits in two.

Apache Gravitino 1.3 takes the other path: govern data where it lives. One catalog can serve multiple clouds at once. A Gravitino Iceberg REST Catalog can federate other IRC services without copying a single byte of metadata. The Glue catalog that came with your AWS account gets governed in place. And identity now works out of the box.

Join the Datastrato team (the original creator of Gravitino) for a deep dive into what's new in 1.3 and what it means for how you operate your data estate. Speakers:

What we'll cover

  • One catalog, every cloud. Multi-backend dispatch in the Iceberg REST Catalog, with short-lived credentials minted per request and refreshed across S3, GCS, OSS, and ADLS. The difference between clouds comes down to a LOCATION clause.

  • IRC federation. Register a remote catalog by name, copy no metadata, and let the owning catalog keep its own authorization, its own credentials, and its own audit log. Federation works across clouds and across on-prem boundaries.

  • AWS Glue, governed in place. Bring Glue estates under one governance model with no migration, queried through Trino and Spark with the same roles, privileges, and audit trail as everything else.

  • Enterprise identity, built in. A local identity provider, break-glass accounts for when your external IdP is down, group-aware ownership, and role inheritance.

  • The table metadata cache hits GA. Enabled by default, reads measured 4x faster under concurrent load, with the access check and a staleness guardrail enforced on every hit.

Plus first-class views, hierarchical namespaces, and a look at what's coming in Datastrato Enterprise 1.3. We'll leave plenty of room for questions.

Who should come

This one is built for the Gravitino community: contributors, maintainers, and anyone running it in production. If you've been tracking the roadmap or weighing the upgrade, bring your questions.

Avatar for Data For AI Events
Presented by
Data For AI Events
Hosted By