API 페이지 매김·필터링·정렬

An endpoint that returns 50,000 rows is slow, heavy, and crashes clients. Pagination returns data in manageable chunks, while filtering and sorting let clients ask for exactly what they need.

The simplest approach uses page and limit. You skip (page - 1) * limit rows and take limit rows.

Prisma exposes skip and take for offset pagination.

Clients need to know how many pages exist. Return the data plus a total count and computed page info.

Large offsets are slow because the database still scans skipped rows, and inserts can shift items between pages. For big or live datasets, prefer cursor pagination.

A cursor points to the last item seen. The next request fetches rows after that cursor — fast and stable even as data changes.

Read filter params from the URL and build a where clause. Use contains for search-style filters.

Build the where object conditionally so only provided params apply. Prisma combines them with AND by default.

Accept a sort field and direction, but whitelist allowed fields so clients cannot order by arbitrary or sensitive columns.

Coerce and clamp inputs: cap limit so a client cannot request a million rows, and default page to 1.

Design scalable list endpoints:

• Always paginate large collections
• Use cursor pagination for big or live data
• Return metadata (total, pages)
• Whitelist sort fields and clamp limits

Test your pagination knowledge.

You learned to build scalable list APIs:

• Paginate with skip/take or cursors
• Return total and page metadata
• Build dynamic where clauses for filtering
• Whitelist sort fields and clamp the limit

Your API now handles large datasets efficiently.