Z. Li and K. Ross. Fast Joins Using Join Indices. The VLDB Journal, 8(1):1--24, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....nested sets in object oriented and object relational database systems. In contrast, OHJs work for any kind of equi join, OHJs are order preserving (not just nested set preserving) and applicable in pure relational as well as object oriented and object relational database systems. Li and Ross [LR99] describe two new join algorithms that are based on join indices. One of the algorithms is sort based, the other is partition based. Both algorithms have in common, that they draw profit from not completely materializing the join result, rather they store the temporary result on disk in two ....

Z. Li and K. A. Ross. Fast joins using join indices. The VLDB Journal, 8(1):1--24, May 1999.

....execution is often I O bounded, without deploying multiple I O channels and disks, simply increasing the number of parallelly executing processors will only have a marginal effect on the performance of join execution when the number of processors reach a certain level. Join index was proposed [4, 16] to speed up join execution. It is essentially a precomputed join to turn a join between two relations R and S into a selection in a join index. Join index is a two attribute relation. Each tuple of a join index has two entries. One entry is a pointer to a tuple in R and the other entry is a ....

Z. Li, K. A. Ross. Fast Joins Using Join Indices. Technical Report, CUCS-032-96, 1996, Columbia University.

....database architectures and join techniques have been proposed in recent years which hold hopes to further improve the performance of join execution in relational database. These techniques include signature method [15, 31, 34] clustering of data tuples and partition of relations [59] join index [28, 61], composite index [33] layered relational database [52] etc. Essentially, the improvements these new techniques made are either better organizing of physical storage of data on the disk to take advantage of some special cases such as sequential disk accesses, or adding some new clever indexing ....

Z. Li, K. A. Ross. Fast Joins Using Join Indices. Technical Report, CUCS-032-96, 1996, Columbia University.

....(a, 3.2) h, 2.2) i, 1.1) a 3.2 b 1.2 c 2.1 d 4.1 e 3.1 f 4.2 g 5.1 h 2.2 i 1.1 (a) B tree (b) Hash Table (c) Direct Mapping Fig. 2. Mapping techniques indices is reported in [BK89, KM90, XH94] A recent paper describes two new join algorithms that are based on join indices [LR99] Both algorithms store the join result in two files on disk, which need to be merged to obtain the actual join result. Therefore, their algorithms achieve their performance gains mostly in situations where the join needs to be materialized and less in situations where the join result is further ....

Li Z, Ross KA (1999) Fast joins using join indices. VLDB J 8(1): 1--24

....and work on path indices is reported in [BK89,KM90,XH94] A join index materializes the join between two relations, say R and S, by storing all the RID pairs of those tuples r, s that satisfy the join predicate. A recent paper describes two new join algorithms that are based on join indices [LR99] One of these algorithms is based on sorting, the other is based on partitioning. Both algorithms have in common that they store the join result on disk such that two files are created: one with R tuples and one with S tuples which need to be merged to obtain the join result. Therefore, their ....

Z. Li and K. A. Ross. Fast joins using join indices. The VLDB Journal, 8(1):1--24, May 1999.

....subquery between iterations, and to reevaluate just the variant part on each iteration. Integrating this technique into a cost based optimizer required careful design. Our work was implemented in the Sybase IQ commercial database system, and will be present in their next commercial release. In [4] we propose techniques for performing a join of two large relations using a join index. Our techniques are novel in that they require only a single pass of each participating relation, even if both tables are much larger than main memory, with all intermediate I O performed on tuple identifiers. ....

Z. Li and K. Ross. Fast joins using join indices. VLDB Journal, 1998. (to appear).

....observe and correct performance bugs in the implementation, while our implementation also highlighted inaccuracies in the cost model, which we were then able to improve. 8 Extensions We outline below several extensions to Jive join and Slam join. More detail about these extensions can be found in [Li and Ross, 1996]. Multi level Recursion When the inputs are larger than the memory bound (Equation 3) it is possible to perform an additional partitioning step (with an associated extra I O cost) to subsequently apply Slam join or Jive join. Tapes Jive join and Slam join are particularly well suited to joining ....

Li, Z. and Ross, K. (1996). Fast joins using join indices. Technical Report CUCS-032-96, Columbia University.

....and analyzed a join algorithm that uses the join index. The most important conclusion of that study was that, under many circumstances, having the join index allows one to compute the join significantly faster than the best ad hoc methods such as Hybrid hashjoin [6] However, it was shown in [12] that Valduriez s algorithm utilizes a significant amount of repetitious I O. Blocks are accessed often for only a small fraction of their tuples. The same block may be read multiple times on different passes within the algorithm. In [12] Li and Ross proposed two algorithms that significantly ....

....such as Hybrid hashjoin [6] However, it was shown in [12] that Valduriez s algorithm utilizes a significant amount of repetitious I O. Blocks are accessed often for only a small fraction of their tuples. The same block may be read multiple times on different passes within the algorithm. In [12], Li and Ross proposed two algorithms that significantly improve upon Valduriez s algorithm. The algorithms are called Jive join and Slamjoin. The two algorithms are duals of one another, and have very similar performance. Jive join range partitions the tuple ids of the second input relation, ....

[Article contains additional citation context not shown here]

Z. Li and K. Ross. Fast joins using join indices. Technical Report CUCS-03296, Columbia University, 1996.

No context found.

Z. Li and K. Ross. Fast Joins Using Join Indices. The VLDB Journal, 8(1):1--24, 1999.

No context found.

Z. Li and K.A. Ross. Fast Joins Using Join Indices. VLDB Journal, 8(1), 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC