Llm Gateway Cost And Caching easy
A retail assistant uses a semantic cache. Weeks after a store changed its holiday hours, users still receive the old hours from cached answers. What is the right fix?
- A Raise the similarity threshold so only exact repeats of a question reuse the answer.
- B Partition the cache by tenant so each store's answers are kept in a separate space.
- C Increase the cache size so newer answers push the older entries out of it sooner.
- D Give entries a TTL plus invalidation when source data changes.
Answer: D. Give entries a TTL plus invalidation when source data changes.
Stale answers are a **freshness** problem, not a matching problem: the query matched correctly, but the cached content was outdated. A TTL plus invalidation trades some hit rate for correctness. A tighter threshold would still serve the stale answer to exact repeats, and tenant partitioning fixes leaks between tenants, not staleness.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- Your primary LLM provider is fully down for an hour, and no tested fallback exists for your agent feature. The product also has an FAQ answer feature backed by a semantic cache. What is the best degradation mode?
- Your gateway applies one 5-second timeout to all LLM calls. Classification calls work fine, but long streamed answers are frequently cut off. What is the best fix?
- A multi-tenant B2B support assistant answers questions using each customer's private knowledge base. To raise the semantic cache hit rate from 22% to 35%, an engineer proposes one shared cache across all tenants. What is the main problem?
- During a provider brownout returning 429 and 503 errors, the client SDK retries 3 times and the gateway also retries 3 times, both immediately. Provider load spikes, and the bill grows from partially streamed responses. What should you change?
- Finance asks for LLM spend broken down by tenant and by product feature. Today the gateway only sees the total on the provider's monthly invoice. What should you do?
- A feature sends 1,500 input tokens and receives 600 output tokens per request. Output tokens cost about 5x input tokens. Which single change cuts cost the most?