Ming-Ling Lo and Chinya V. Ravishankar. Towards Eliminating Random I/O in Hash Joins. IEEE. Proc. of 12th Int'l Conf. on Data Engineering, 1996.  Home/Search   Document Details and Download   Summary   Related Articles

....attribute values can be used to filter out unmatchable tuples of S while scan S. This is a similar but more generalized method than semijoin. Other more general methods of semijoins can be found in [18, 26] Eliminating random I O: Hash join has two phases: partitioning and joining phases. In [62], disk I Os are divided into four phases: PR, PW, JR, JW. PR represents partitioning phase read, PW represents partitioning phase write, JR represents joining phase read, and JW represents joining phase write. For convenience of comparison, in this paper we assume PW and JR are sequential. ....

....write, JR represents joining phase read, and JW represents joining phase write. For convenience of comparison, in this paper we assume PW and JR are sequential. Actually, this assumption is not totally true. The proposed optimization targets reducing random disk I Os on two phases: PW and JR. In [62], hash buckets are comprised of bucket segments. Each bucket segment is a physically contiguous storage area. Several bucket segments can form a write group. 23 The memory for performing join operation are divided into three parts: IB, the input buffer, SB, the staging buffer, and OB, the output ....

[Article contains additional citation context not shown here]

Ming-Ling Lo and Chinya V. Ravishankar. Towards Eliminating Random I/O in Hash Joins. IEEE. Proc. of 12th Int'l Conf. on Data Engineering, 1996.

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