B-Trees and B Plus Trees in Data Management: A Comparative Analysis of Search and Range-Query Performance

Database indexes help a system find records without scanning an entire table. B-Trees and B Plus Trees both organize ordered data, but they store entries differently: a B-Tree may keep entries in internal nodes as well as leaves, while a B Plus Tree keeps them at the leaf level and links the leaves in key order. This paper compares the two structures for point searches and range queries. We estimate node capacity from a simplified 4 KB page layout, then evaluate Python implementations across 15 trials. Node visits are counted as a proxy for page accesses, and query time is measured separately. For a range query returning 10000 keys from a tree of 500000 keys, the B Plus Tree recorded 59.27 node accesses on average, compared with 88.21 for the B-Tree. With equal node capacities the counts were nearly the same, which suggests the reduction came mainly from higher fan-out. The B Plus Tree also completed large range scans faster, although the timing difference may partly reflect the traversal code. Point-search times showed no consistent difference. Because the tests ran in memory, the results do not measure physical disk input and output.