Index
• Index is a separate entity
• It is organized similar to data component (CI and CA)
• It has a different CI size
• VSAM builds index when data is loaded
• Index is organized as inverted binary tree
• VSAM compresses key to conserve space
• Can have several levels of indexes
• Lowest level of index is called ‘sequence set’. One sequence set for one CA. These are always in the key order whereas data CI need not be in key order. Each entry has the highest key (or “one” less than the first key of the next CI so that the sequence set need not be updated when higher key added to CI
• Next level is called ‘Index set’. It contains primary keys and pointers to the sequence set. More than one level of index set based on number of sequence set records. Top level index is always one record
CI and CA split
• CI split happens while adding new records or extending an old record and not enough room in the CI to complete the operation
• CI split may trigger a CA split
• Splits generally degrades performance
• Specify free space to reduce CI and CA splits
CI size
• VSAM can select an efficient size if you ignore
• Large CI size for batch sequential processing
• Small CI for random processing
• All are done automatically by VSAM
2 comments for “Index”
Index• Index is a separate entity• It is organized similar to data component (CI and CA)• It has a different CI size• VSAM builds index when data is loaded• Index is organized as inverted binary tree• VSAM compresses key to conserve space• Can have several levels of indexes• Lowest level of index is called ‘sequence set’. One sequence set for one CA. These are always in the key order whereas data CI need not be in key order. Each entry has the highest key (or “one” less than the first key of the next CI so that the sequence set need not be updated when higher key added to CI• Next level is called ‘Index set’. It contains primary keys and pointers to the sequence set. More than one level of index set based on number of sequence set records. Top level index is always one recordCI and CA split• CI split happens while adding new records or extending an old record and not enough room in the CI to complete the operation• CI split may trigger a CA split• Splits generally degrades performance• Specify free space to reduce CI and CA splitsCI size• VSAM can select an efficient size if you ignore• Large CI size for batch sequential processing• Small CI for random processing• All are done automatically by VSAM
+1
