M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. In Proc. VLDB, pp. 134--145, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....structure that can be built on top of an existing spatial indexing techniques, such as an R tree. Consequently, implementing SETI is much easier than implementing a new physical indexing structure. In addition, existing techniques for concurrency control that have been developed for R trees [15, 16] can be directly used by SETI. The remainder of this paper is organized as follows: The problem definition and related work are presented in Section 2. The SETI indexing mechanism is presented in Section 3. Section 4 presents the experimental results based on an actual implementation of SETI, 3D ....

KORNACKER,M.,AND BANKS, D. High-Concurrency Locking in R-trees. In Proceedings of the 21st VLDB Conf. (Zurich, Switzerland, September 1995), pp. 134--145.

....integrator [19] This type of simulation is at the core of solving the protein folding problem, which IBM and others are investigating with tremendous vigor. 3. 2 DIS Data Management The Data Management Benchmark [4] implements a simplified object oriented database with an R Tree indexing scheme [17, 23]. R Trees have the following properties: They are height balanced (ie, all leaves are at the same level) If M is the order of the tree, and k is a constant, every node has between kM and M index entries (except the root) A sub tree may contain a hypercube index if that hypercube can cover the ....

Banks Kornacker. High-Concurrency Locking in R-Tree. In Proceedings of 21st International Conference on Very Large Data Bases, September 1995.

....integrator [19] This type of simulation is at the core of solving the protein folding problem, which IBM and others are investigating with tremendous vigor. 3. 2 DIS Data Management The Data Management Benchmark [4] implements a simplified object oriented database with an R Tree indexing scheme [17, 23]. R Trees have the following properties: They are height balanced (ie, all leaves are at the same level) If M is the order of the tree, and k is a constant, every node has between kM and M index entries (except the root) A sub tree may contain a hypercube index if that hypercube can ....

Banks Kornacker. High-Concurrency Locking in R-Tree. In Proceedings of 21st International Conference on Very Large Data Bases, September 1995.

....and thus poor performance. Many approaches that exploit the structure and the semantics of the operations to provide high concurrency have been developed for B trees [6] These approaches need to be generalized to multidimensional data structures. Initial steps in this direction can be found in [12, 22] which discuss techniques to support concurrent operations on the R tree and the grid file respectively. Second, techniques must be developed to protect the ranges specified in the retrieval from subsequent insertions and deletions before the retrieval commits. Such insertions and deletions are ....

M. Kornacker and D. Banks. High-concurrency locking in r-trees. In Proceedings of Very Large Databases (VLDB), pages 134-145, September 1995.

....the structure and the semantics of the operations to provide high concurrency have been developed for B trees a single dimensional data structure (Gray and Reuter, 1993) These approaches need to be generalized to multidimensional data structures. Initial steps in this direction can be found in (Kornacker and Banks, 1995; Salzberg, 1986) which discuss techniques to support concurrent operations on the R tree and a grid file respectively. Secondly, techniques must be developed to protect the ranges specified in the retrieval from subsequent insertion and deletion before the retrieval commits. Such insertions and ....

Kornacker, M. & Banks, D. (1995), "High-Concurrency Locking in R-Trees", In Very Large Databases (VLDB), pages 134-145, September 1995.

....Postgres, is not public. Many approaches that exploit the structure and the semantics of the operations to provide high concurrency have been developed for B trees [4] These approaches need to be generalized to multidimensional data structures. Initial steps in this direction can be found in [11, 21] which discuss techniques to support concurrent operations on the R tree and the grid file respectively. Second, techniques must be developed to protect the ranges specified in the retrieval from subsequent insertions and deletions before the retrieval commits. Such insertions and deletions are ....

M. Kornacker and D. Banks. High-concurrency locking in r-trees. In Proceedings of Very Large Databases (VLDB), pages 134-145, September 1995.

....the structure and the semantics of the operations to provide high concurrency have been developed for B trees# a single dimensional data structure #Gray and Reuter, 1993#. These approaches need to be generalized to multidimensional data structures. Initial steps in this direction can be found in #Kornacker and Banks, 1995; Salzberg, 1986# which discuss techniques to support concurrent operations on the R tree and a grid #le respectively. Secondly, techniques must be developed to protect the ranges speci#ed in the retrieval from subsequent insertion and deletion before the retrieval commits. Such insertions and ....

Kornacker, M. & Banks, D. #1995#, #High-Concurrency Locking in R-Trees", In Very Large Databases #VLDB#, pages 134-145, September 1995.

....and dynamic declustering techniques including mapping functions, local load balancing and maxcut graph partitioning; parallelizing spatial indexes, range queries[16] spatial join[17] and spatial analysis; Distributed architectures such as GIS T. 10. Trends and Open Issues: Concurrency Control[18], Digital library, quality assurance for spatial data, collection and maintenance of spatial data, headup digitization, archival global positioning systems and mobile GIS, spatio temporal modeling, imprecise boundaries, multi scale generalization, data lineage etc. ....

M. Kornacker and D. Banks, High-Concurrency Locking in R-trees, Proc. VLDB Conf., (1995).

....of cost models for query strategies, and the development of new spatial join algorithms beyond nested loop and tree matching. Many of the research needs identified in [15] have since been addressed. For example, concurrency control techniques for R trees have been studied in the context of R link [16] trees. Also, new spatial join strategies using space partitioning [22] have been explored. In this paper, we identify the recent accomplishments in spatial databases as well as current research needs, based on publications in journals and conference proceedings and recent commercial trends. 1.3 ....

....to disk pages. Many variations of the R tree structure exist whose main emphasis is on discovering new strategies to maintain the balance of the tree, in case of a split, and to minimize the overlap of the MBRs in order to improve the search time. Concurrency control for spatial access methods [16] is provided by the R link tree, which is a variant of the R tree with additional sibling pointers that allow the tracking of modifications. Concurrency is provided during operations such as search, insert, and delete. The R link tree is also recoverable in a write ahead logging environment. ....

[Article contains additional citation context not shown here]

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. In Proceedings of 21st International Conference on Very Large Data Bases(VLDB'95), pages 134--145, Zurich, Switzerland, September 1995.

....solution to concurrency problems in a treebased index structure is to lock the entire tree or the subtree that needs to be modified. The upper levels of the subtree are locked so that only readers can access them [2] To achieve much higher levels of concurrency, B link trees [18] and Rlink trees [15] were proposed, and Kornacker et al. 16] generalized the ideas of R link trees to apply to a broader class of tree based access methods. Using the link based concurrency control protocols, a lock on a parent node can be released before visiting a child node. To use this protocol in an ....

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. Proceedings of VLDB, pp. 134--145 (1995).

....in a tree based index structure is to lock the entire tree or the subtree that needs to be modified during insertions. The upper levels of the subtree are locked so that only readers can still access them [BS77] To achieve much higher levels of concurrency, B link trees [LY81] and R link trees [KB95] were proposed, and Kornacker et al. KMH97] generalized the ideas of R link trees to apply to a broader class of tree based access methods. Using the link based concurrency control protocols, a lock on a parent node can be released before visiting a child node. To use this protocol in an ....

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. Proceedings of VLDB, pp. 134-- 145 (1995).

....locking and without affecting query behavior. The total ordering of data items is implicitly used for ensuring such correct query behavior. Unfortunately, such an ordering does not exist in multiple dimensions. Therefore, for multi dimensional structures such as R trees [4] Kornacker and Banks [7] associate sequence numbers with child entries in an index node and use the total ordering on these numbers to ensure correct query behavior. The extra space for these sequence numbers reduces the capacity of tree nodes, which may lead to a degradation in query performance. Consequently, Kornacker ....

M. Kornacker and D. Banks. High concurrency locking for R-trees. Proceedings of the Int. Conf. on Very Large Data Bases, pages 134--145, September 1996.

....specialized internal structure required by hB trees. ffl Concurrency Control, Recovery and Consistency: High concurrency, recoverability, and degree 3 consistency are critical factors in a full fledged database system. We are considering extending the results of Kornacker and Banks for R trees [KB95] to our implementation of GiSTs. ffl Variable Length Keys: It is often useful to allow keys to vary in length, particularly given the Compress method available in GiSTs. This requires particular care in implementation of tree methods like Insert and Split. ffl Bulk Loading: In unordereddomains, ....

Marcel Kornacker and Douglas Banks. HighConcurrency Locking in R-Trees. In Proc. 21st International Conference on Very Large Data Bases, Zurich, September 1995.

....Lehman and Yao s sideways pointer techniques, along with recovery techniques, to R trees. Ng and Kameda [NK94] do so by generating a pending update list at each node of the tree, and applying Lomet and Salzberg s results on Pi trees to this context. Banks, Kornacker and Stonebraker [BKS94, KB95] have a simpler solution that marks keys and nodes with sequence numbers, and uses the sequence numbers to determine order among the nodes. Both approaches provide solutions for degree 3 consistency Ng and Kameda via aborting transactions that read phantoms, and Banks, et al. via predicate ....

Marcel Kornacker and Douglas Banks. High-Concurrency Locking in R-Trees. Submitted for publication, 1995.

....structure modifications have been recognized early on and have been published in a number of articles [ML92, Moh90a, GR93, LS92] The GiST logging and recovery protocol as presented in Section 4.4 directly builds on that prior work. The basis of the GiST concurrency protocol was developed in [KB95] in the context of R trees, which have the same structural properties as GiSTs (non partitioning, non linear keys) The R tree concurrency control protocol replicates the NSN of a node in its parent entry, which adds extra space overhead. The paper also does not sufficiently address the problems ....

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. In Proc. 21st Int'l Conference on Very Large Databases (VLDB), Z urich, Switzerland, pages 134--145, September 1995.

....exist. Figure 1: Incorrect Interleaving of Key Search and Node Split operation if a node has split and if a right sibling might contain entries intersecting the search range. GiSTs do not impose these restrictions on the key domain, which means that the B link strategy by itself is insufficient. [KB95] adapts the link strategy to R trees by assigning sequencenumbers to the nodesand incrementing these during node splits. The sequence number is also recorded in the parent entry, which allows traversing operations to reconstruct the lineage of a node after it was split. In contrast to B link ....

....on that prior work. The basic GiST structure itself is described in [HNP95] The paper also gives three examples of implementations of specific trees within the GiST structure and tries to analyze the GiST performance in a general way. The basis of the GiST concurrency protocol was developed in [KB95] in the context of R trees, which have the same structural properties as GiSTs (non partitioning, non linear keys) The paper does not sufficiently address the problems of transactional isolation, recovery and node deletion. The data only locking approach and logical deletion have beenadoptedfrom ....

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. In Proc. 21st Int'l Conference on Very Large Databases (VLDB), pages 134--145, September 1995.

No context found.

M. Kornacker and D. Banks. High-Concurrency Locking in R-Trees. In Proc. VLDB, pp. 134--145, 1995.

No context found.

M. Kornacker and D. Banks: "High-Concurrency Locking in R-trees", Proceedings 21st VLDB Conference, pp.134-145, Zurich, Switzerland, 1995.

No context found.

Kornacker M., Banks D.: High-Concurrency Locking in R-Trees. Proc. 21st Int. Conf. on Very Large Databases (VLDB): 134-145, 1995.

No context found.

Kornacker M., Banks D.:High-Concurrency Locking in R-Trees . Proc. 21st Int. Conf. on Very Large Databases (VLDB): 134-145, 1995.

No context found.

Banks Kornacker. High-Concurrency Locking in RTree. In Proceedings of 21st International Conference on Very Large Data Bases, September 1995.

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