Phase 2: reducing rows read without Redis

Last month we reached 3,000,000,000+ rows read, so this phase focuses on reducing database scans in the most frequently requested race endpoints.

What changed in the endpoints

We refactored the main season/round and current/last endpoints to avoid large multi-table joins as the first step.

Before:

Big joins across session + races + championships + circuits + drivers + teams in a single query.
Filters like year and round were applied after joining big datasets.

Now:

Step 1 query: resolve the target race with indexed filters (championshipId + round or latest session date).
Step 2 query: fetch only the session rows for that raceId and then join lightweight dimensions (drivers, teams).
Reduced joins by removing Championships where we already have championshipId in Races.

Updated routes:

We kept response caching at the HTTP layer:

This gives lower database pressure for repeated requests while keeping the API fresh enough for race data use cases.

Middleware rate limiter now fails safely if Upstash environment values are missing or invalid.
Blog page layout now uses min-h-screen instead of a fixed h-screen, preventing article-list collapse while scrolling.
Fix error #43

We are always open to get PRs, translations, issues, articles and suggestions from the community in Github.

F1 Connect it´s a free and open source project, so we aprecciate a lot any Github star or sharing our work with the community.

Thanks from the F1 Api team.