Which approach can fetch alternate rows from a table?

• A. RANK() with a modulo filter to select alternate rows
• B. ROW_NUMBER() with a modulo filter to select alternate rows
• C. NTILE() with a modulo filter to select alternate rows
• D. DENSE_RANK() with a modulo filter to select alternate rows

Answer: (B)

To fetch alternate rows reliably, you need a stable, per-row sequence in a defined order. ROW_NUMBER() provides that by assigning a unique, consecutive number to every row according to your ORDER BY clause. Once every row has a unique index, you can simply filter where the index modulo 2 is 0 (or 1) to get every other row.

Why this works best: ROW_NUMBER() guarantees a strict 1, 2, 3, ... sequence with no ties, so the alternation pattern is predictable and reproducible. Other approaches can introduce ties or groupings that break a clean alternation: RANK() and DENSE_RANK() assign the same value to tied rows, which means several rows can share a rank and the sequence isn’t strictly alternating. NTILE() divides the result into a fixed number of groups rather than giving a per-row linear index, so filtering by group number won’t produce a simple alternating pattern in the overall order.

So, using ROW_NUMBER() with a modulo filter on its result is the straightforward, reliable way to fetch alternate rows, as long as you define a deterministic order in the window.