Table 1 On-line tuning of row-stores, a summary
Study | Structures | Experiments | Notes |
|---|---|---|---|
COLT [35] | Indexes | PostgreSQL | Uses storage budget constraint; employs “what-if” optimizer mode; manages active, hot and cold sets of indexes |
Alerter [36] | Indexes, materialized views | SQL server | Uses storage budget and mininum improvement constraints; Notifies DBA and provides a set of candidate structures |
Alerter [49] | Aggregation tables | Mondrian | Uses soft storage budget constraint |
AdaptPD [50] | Vertical partitioning | SQL server | Employs “what-if” optimizer mode with caching |
WFIT [51] | Indexes | IBM DB2 Express-C | Takes a workload and user feedback into account |
Materialized views, storage selection | Multiple | Tuning of multistore system physical design. Uses storage and transfer constraints | |
EVO [54] | Indexes | Multiple | Authors propose query plan transformations using genetic algorithm for index selection |
ARH [55] | Automatic re-indexing | PostgreSQL | A set of heuristics is used to decide when to trigger a re-indexing process to counter an index fragmentation |
Tuner [33] | Multiple | PostgreSQL | An on-line tool which tunes several physical structures—indexes, partitions, and is capable of tracking index interaction |
AutoStore [56] | Vertical partitioning | Custom | A comparison of on-line algorithms for vertical partitioning |
SMOPD [57] | Vertical partitioning | Custom | Closed itemset mining for on-line vertical paritioning |