During peak times Cloud Storage reads return HTTP 429 Too Many Requests. How should your image-processing service handle this?

The correct approach is to retry the request with truncated exponential backoff. HTTP 429 errors indicate that a service is temporarily overloaded or that you've exceeded a rate limit. Implementing exponential backoff involves waiting for increasingly longer periods between retries, which helps to alleviate the load on the service and increases the likelihood of success. "Truncated" means there's an upper limit to the backoff duration, preventing excessively long waits. Adding a cache-control header (incorrect option) primarily affects how browsers or proxies cache content, not how the Cloud Storage service handles read requests under heavy load. Requesting a quota increase (incorrect option) is a long-term solution for sustained high usage, but it doesn't immediately address transient 429 errors during peak times. Changing the bucket storage class to Multi-Regional (incorrect option) affects data availability and latency across regions, but it doesn't inherently prevent or resolve 429 errors caused by high request rates to a specific bucket.