Problem description
A high volume of requests targeting the same key prefix simultaneously causes OSS to throttle those requests and return a 503 PartitionQpsLimitted error.
Causes
OSS partitions data based on the UTF-8 encoding order of object keys. When object keys share a sequential prefix—such as a date-based timestamp or a string in lexicographic order—writes and reads cluster into the same partition, creating a hot spot. Once the hot spot partition reaches its request rate limit, OSS applies server-side throttling.
Examples
The following sequential object keys all map to the same partition range:
sample-bucket-01/2024-07-19/customer-1/file1
sample-bucket-01/2024-07-19/customer-2/file2
sample-bucket-01/2024-07-19/customer-3/file3
sample-bucket-01/2024-07-20/customer-2/file4
sample-bucket-01/2024-07-20/customer-5/file5
sample-bucket-01/2024-07-20/customer-7/file6Sending a high volume of concurrent requests to these keys overloads the partition and triggers the 503 PartitionQpsLimitted error.
Solutions
Distribute requests across multiple partitions by randomizing object key prefixes so that concurrently accessed objects do not share the same prefix.
For broader guidance on designing object key naming strategies to maximize throughput, see OSS performance best practices.