The index page cluster ratio is used to estimate the cost of large I/O for queries that need to read a large number of leaf-level pages from nonclustered indexes or clustered indexes on data-only-locked tables. Some examples of such queries are covered index scans and range queries that read a large number of rows.
On newly created indexes, the “Index page cluster ratio” is 1.0, or very close to 1.0, indicating optimal clustering of index leaf pages on extents. As index pages are split and new pages are allocated from additional extents, the ratio drops. A very low percentage could indicate that dropping and re-creating the index or running reorg rebuild on the index would improve performance, especially if many queries perform covered scans.
See “How cluster ratios affect large I/O estimates” on page 67 in the Performance and Tuning: Optimizer for more information.
