152 CHAPTER 8 Multidimensional Clustering
– Using MDC data is never placed in the wrong cell.
• Scans on any dimension index provide clustered data access.
– Each block ID identifies a storage block guaranteed to contain data having
that dimension value.
• Dimensions/slices can be accessed independently using their block indexes.
– One block index does not impact the cluster factor of another block index.
• Block index scans can be combined (ANDed, ORed).
– The resulting list of blocks also provides clustered data access.
• Access to the clustered data is much faster.
– More efficient scan using a block ID to access blocks versus multiple RIDs
for each row.
• Enabled MDC-specific ‘block lookahead prefetching’.
– Read ahead in the index and prefetch the ...