Question

An e-commerce application stores shopping cart documents in an Azure Cosmos DB for NoSQL container. Multiple client sessions can update the same cart concurrently. The development team must ensure that when two users update the same cart at nearly the same time, one update does not silently overwrite the other, without introducing pessimistic locking that would reduce throughput. Which approach should the team implement?

Accepted Answer

Read the item, capture its _etag, and include an If-Match header with that value on the replace operation, retrying if a 412 Precondition Failed is returned.

Answer

Set a Time to Live (TTL) on the container so stale cart documents automatically expire before conflicting writes occur.

Answer

Use a transactional batch that groups the read and write operations for the cart into a single atomic call.

Answer

Enable the strong consistency level on the account so all writes are serialized and conflicts cannot happen.