Problem

At Nimbus Storage, block requests are processed in order, but the retrieval layer can prefetch a limited number of distinct blocks at a time. Given a sequence of requested block IDs and a cache capacity k, write a function that returns the length of the longest contiguous request segment that can be served using at most k distinct block IDs.

This models optimizing retrieval by maximizing the largest request window that fits within the storage system's active block set.

Formal Specification

• Input:
  • requests: a list of integers representing block IDs
  • k: an integer representing the maximum number of distinct block IDs allowed in the active retrieval window
• Output:
  • An integer: the maximum length of a contiguous subarray containing at most k distinct values

Examples

Example 1

• Input: requests = [1, 2, 1, 2, 3], k = 2
• Output: 4
• Explanation: The longest valid segment is [1, 2, 1, 2], which contains only block IDs 1 and 2.

Example 2

• Input: requests = [4, 4, 4, 4], k = 1
• Output: 4
• Explanation: All requests are for the same block, so the entire array is valid.

Constraints

• 1 <= len(requests) <= 10^5
• 0 <= requests[i] <= 10^9
• 0 <= k <= len(requests)
• The solution should run in linear time.