Academic
Publications
Efficient Query Processing for Multi-Dimensionally Clustered Tables in DB2

Efficient Query Processing for Multi-Dimensionally Clustered Tables in DB2,Bishwaranjan Bhattacharjee,Sriram Padmanabhan,Timothy Malkemus,Tony Lai,Les

Efficient Query Processing for Multi-Dimensionally Clustered Tables in DB2   (Citations: 12)
BibTex | RIS | RefWorks Download
We have introduced a Multi-Dimensional Clustering (MDC) physical layout scheme in DB2 version 8.0 for relational tables. Multi- Dimensional Clustering is based on the def- inition of one or more orthogonal clustering attributes (or expressions) of a table. The table is organized physically by associating records with similar values for the dimension attributes in a cluster. Each clustering key is allocated one or more blocks of physical storage with the aim of storing the multiple records belonging to the cluster in almost con- tiguous fashion. Block oriented indexes are created to access these blocks. In this pa- per, we describe novel techniques for query processing operations that provide signican t performance improvements for MDC tables. Current database systems employ a repertoire of access methods including table scans, index scans, index ANDing, and index ORing. We have extended these access methods for e- ciently processing the block based MDC ta- bles. One important concept at the core of processing MDC tables is the block oriented access technique. In addition, since MDC ta- bles can include regular record oriented in- dexes, we employ novel techniques to combine block and record indexes. Block oriented pro- cessing is extended to nested loop joins and
Conference: Very Large Data Bases - VLDB , pp. 963-974, 2003
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...An example of advanced database clustering can be found in [2], wherein data rows are partitioned with regards to their values on multiple columns...

    Dominik Slezaket al. Intelligent Data Granulation on Load: Improving Infobright's Knowledge...

    • ...Partitions Fig.1. Partitioning a two-column table...
    • ...Figure 1 illustrates this on a two-column table...
    • ...We recursively split thekthpartition from thek -1 partition atallplaces inthefrequency sorted list ofofvalues that are compatible withtherulesabove...
    • ...We thencall theoptimization steprecursively tofindtheoptimal partitioning oftheremaining values intok-1 partitions...
    • ...We compute 4 masks(once percell): 1/Masktoextract needed fields...
    • ...1.Forallcellswithina drawer, thefieldcodes for eachgrouping columnaredrawnfroma single dictionary...
    • ...1)Variability withQuerySelectivity. Ournextexperiment concerns theperformance ofqueries that haveselective predicates...
    • ...Incontrast tomulti-dimensional clustered indexes (eg, [1]), frequency partitions arechosen to optimize forcompression, notforfiltering...

    Vijayshankar RamanOet al. Constant-Time Query Processing

    • ...More recently, DB2 introduces a new physical layout which is more efficient than regular clustering indexes, named Multi-Dimensional Clustering (MDC) [24]...
    • ...More details of the MDC table can be found in [24] and DB2 online help center...

    Li Maet al. Effective and efficient semantic web data management over DB2

    • ...More recently, Multi-Dimensional Clustering (MDC) [12] is developed to support block indexes which is more efficient than normal clustering indexes...

    Jing Luet al. SOR: A Practical System for Ontology Storage, Reasoning and Search

    • ...Examples include Multi Dimensional Clustering (MDC) [1][2] in DB2 UDB [9], Partitioned Primary Index in Teradata [3] and Index Clustered Tables in Oracle [4]...
    • ...Block index scans are used very frequently on MDC tables [1][2] in decision support systems that require slicing and dicing of large datasets...
    • ...While the presented algorithms work for any type of index-based scans, we have decided to implement our prototype first for MDC block index scans [1] [2] since these are increasingly popular in customer settings...

    Christian A. Langet al. Increasing Buffer-Locality for Multiple Index Based Scans through Inte...

Sort by: