How does Uber scale to millions of concurrent requests?

Uber senior staff engineer Ankit Srivasta and Cloud Spanner engineering director Preetham Shah describe Uber Fulfillment's 2022 migration from a Cassandra-based platform to Spanner.

Key Points Covered

• Ankit explains why cross-row and cross-table inconsistencies made consistency a primary datastore requirement. [00:03:16]-[00:06:25]
• The hybrid design used redundant routes, private API routing, and route benchmarks. [00:06:25]-[00:08:31]
• Active orders remained on their original stack while safe sessions moved progressively through test environments and bounded rollouts. [00:09:34]-[00:11:43]
• The teams subsequently tuned observability, failover, autoscaling, read routing, and caching. [00:11:43]-[00:18:03]
• This 2022 first-party account is not a reusable reference design; check current documentation and validate workload, consistency, failure, security, data, and regulatory behavior independently.

Full video: https://www.youtube.com/watch?v=DY2AR8Wzg3Y(opens in a new tab)