You need to ensure that a Cosmos DB container's throughput is never overwhelmed by a single tenant in a multi-tenant app. Which approach best isolates this risk?
-
A
Use a shared database with autoscale RU/s and per-tenant partition keys
-
B
Create a dedicated container per tenant with provisioned RU/s
-
C
Use serverless Cosmos DB with no RU limit
-
D
Enable geo-replication to distribute load