Case Study: eBay achieves mission-critical multi-data-center availability and resilience with MongoDB

eBay Building Mission-Critical Multi-Data Center Applications with MongoDB

eBay, a top-10 global retail brand with 170+ million active buyers and 1 billion live listings, must deliver always-on customer experiences across 190 markets. To avoid costly downtime on commodity hardware and meet global availability needs, eBay’s DBA team created “Resiliency Design Patterns” to guide database selection and architecture, balancing availability, consistency, durability, recoverability, scalability, and performance across more than 3,000 non-relational instances and multiple petabytes of data.

The team standardized on MongoDB patterns that distribute replica sets and sharded clusters across multiple data centers—e.g., a 7-node replica set with election priorities across three US data centers, a 50-member read-scaled replica set for the product catalog, and sharded clusters for high read/write workloads—tuned with write/read concerns and zone sharding. These designs maintain availability during data-center failures, reduce recovery impact, allow MongoDB to serve system-of-record use cases, and scale to support eBay’s massive, customer-facing applications while newer features (zone sharding, retryable writes) further improve multi-DC write availability and simplify application logic.

eBay

Feng Qu

eBay