Design split-stay combinations API
Overview
The question evaluates API and system design competence, including data modeling for availability intervals, request/response schema design, deterministic pagination and ordering, deduplication of unordered listing pairs, validation/error handling, and performance considerations such as indexing, precomputation, caching, and SLAs.
Design an HTTP API for Two-Listing Split-Stay Search
Context
You are designing search for a travel marketplace that supports split stays across two listings. Inventory is represented per listing as an array of day numbers (integers) indicating available nights. Given a target date range [S, E] (inclusive, in day numbers), return all two-listing combinations that can cover the range with consecutive subsegments: the first listing covers S..T and the second covers T+1..E for some split point T where both subsegments are non-empty.
Assume day numbers correspond to UTC-normalized yyyy-mm-dd dates mapped to an integer day index (e.g., days since epoch). If you prefer ISO dates in the API, define the mapping internally.
Requirements
Design a single HTTP API endpoint that:
- Request/response schemas
- Define how listings are identified.
- Include parameters for date range [S, E].
- Include optional filters (e.g., minimum stay lengths per segment, distinct listings only).
- Define the response format that returns split-stay combinations and optionally details about valid split points.
- Pagination and ordering
- Define deterministic ordering of results.
- Define pagination (limit and cursor/offset) and stability guarantees.
- Deduplication
- Ensure unordered pairs are deduplicated when multiple split points T produce the same two listings. Return each unordered pair at most once, with an optional list of split points.
- Validation and error handling
- Specify parameter validation rules and error responses.
- Performance and system design
- Define performance targets (latency, throughput, scale limits) and SLAs.
- Discuss indexing/precomputing maximal consecutive intervals and how they are queried.
- Outline the query flow and algorithm to find valid pairs.
- Discuss caching, consistency model, idempotency, rate limiting, and observability.
- Sample correctness
-
Show how the endpoint would produce [["B","C"]] for:
- A: [1,2,3,6,7,10,11]
- B: [3,4,5,6,8,9,10,13]
- C: [7,8,9,10,11]
- date range [3,11]
State any minimal assumptions you need. Keep the spec precise and implementable.
Solution
Login required