95% lower egress costs on a hybrid, multi-region lakehouse
A US-headquartered bank serving 90 million+ customers unified analytics across AWS, Azure, and on-prem storage. Data residency rules stayed intact across 150+ countries, and the queries got faster.
"We have been impressed with e6data's locality-aware architecture and its ability to handle our multi-region workloads with low cost, while maintaining the SLAs."
One lakehouse, every region, no copies
Compute went to the data instead of data going to the compute. The residency rules never moved, and neither did the tables.
The bind
Petabytes siloed across AWS in the US, Azure in Europe, and on-prem storage in Asia, walled in by residency law. Cross-region joins averaged ~5 seconds at p95 against a sub-second SLA from trading desks, and every morning terabytes were copied between regions just so queries could run.
The hidden bill
Over 50 TB a month crossed regions at $0.02 per GB. Keeping the copies in sync consumed 6+ Spark pipelines and hundreds of engineering hours, and quarterly audits still flagged ~30% of tables as out of policy because copies drifted from the Unity catalog.
The evaluation
Before any rollout, the bank set the bar: every control and data-residency rule had to stay in force, cross-border egress had to fall, and locality-aware execution had to be proven on their own data and scored against the incumbent, not a synthetic benchmark.
The drop-in
e6data's hybrid lakehouse deployed on the existing infrastructure. Locality-aware execution runs partial plans inside each region and merges only minimal aggregates across borders. Zero-copy pushdown queries S3, Blob, and on-prem stores directly, and Unity Catalog policies extend to every environment with central logging.
The result
After a 6-week pilot: egress down 84%, per-query network cost down 95 to 99%, p95 down to ~1.0 second, and zero policy exceptions at audit. One unified lakehouse now serves all regions without maintaining data copies or custom pipelines.
Book a demo on your own workloads
Reach out to book a demo, share challenges you're facing, and tell us how this fits into what you're currently working on or thinking about.
Prefer to self-serve? Problems we're solving →