T. Johnson and D. Shasha. The Performance of Current B-Tree Algorithms. ACM Transactions on Database Systems, 18(1):51--101, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and an experimental performance analysis, the B link tree is the only well studied relaxed balancing algorithm. There are analytical examinations and simulation experiments which show that the B link tree has a clear advantage over other concurrent but non relaxed balanced B tree algorithms [12, 13, 33]. So far the performance of concurrent relaxed balanced binary search trees has hardly been studied by experiments. Non of the experimental research known to us examine relaxed balanced binary trees in concurrent environments. Boug e et al. perform some experiments in order to study the practical ....

....63 user processes apply concurrently altogether 1000000 random dictionary operations (insertions, deletions, and searches) to the tree. The probabilities that a dictionary operation is an insert, delete, or search operation are p insert = 0:5, p delete = 0:2, and p search = 0:3 (analogously as in [13]) The rebalancing is done by a single rebalancing process that work concurrently with the user processes and gets its work from a problem queue, cf. Section 3.1. The offered lock modes are either all possible modes r, w, and x, or only r and x, or in the third variant only x. The costs of an ....

T. Johnson and D. Shasha. The performance of current B-tree algorithms. ACM Transactions on Database Systems, 18(1):51--101, March 1993.

....studies of concurrency control algorithms, such as [2, 21, 25, 30] which do not assume a graph structured database and adopt a random access pattern model of database workload. Or for example studies of concurrency control algorithms for BTrees in which the database is restricted to a tree [12, 24]. Instead, in our work, we have considered a general graph as a model of the database and access patterns such as breadthfirst and depth first traversal which match well processing requirements for knowledge based systems. Performance of a multi level implementation of the 2PL, with a detailed ....

T. Johnson and D. Shasha. The Performance of Current B-Tree Algorithms. ACM Transactions on Database Systems, 8(1):51--101, March 1993.

....on the number of entries left in the nodes, allowing them to empty completely before simply removing them. According to [GR93] most database system implementations of B trees have adopted this approach. Its most effective use is when it is vital to allow concurrent access to the tree [JS93b], and excessive splitting and merging of nodes would restrict concurrency. JS89] derives some analytic solutions calculating memory utilization for B trees under a mix of insertions and lazy deletions, based on previous research which considered insertions only [BY89] The simulations in ....

T. Johnson and D. Shasha. The performance of current B-tree algorithms. ACM Transactions on Database Systems, 18(1):51--101, 1993.

.... data access [Pap86, BHG87] At the same time, however, applied research has continued to focus, up to this day, on exploiting the particulars of application domains, data structures, or system architectures to further increase the degree of concurrency and to optimize concurrency control protocols [Sag86, ML92, JS93, Lom93]. In practice, a database stores redundant data in various forms to speed up access; data records are inserted and deleted; and transactions perform operations in multiple files and structures. Implementation driven research (not always published, unfortunately) has made progress on the efficient ....

.... (although in practice subsequent invocations should be optimized, because there is a previous reference point, record y) Design of the critical regions that ensure property 1 above depends on the specific structure used to maintain the sequence, and some papers have studied it for various cases [Sag86, SG88, Bil87, JS93]. The choose operation has not been considered explicitly, but it should be relatively easy to implement on top of the results of the referred papers. 3.4 Associative access Arbitrary associative access is the core functionality of query languages, but it is essentially incompatible with the ....

T. Johnson and D. Shasha. The performance of current B-tree algorithms. ACM Transactions on Database Systems, 18(1):51--101, March 1993.

No context found.

T. Johnson and D. Shasha. The Performance of Current B-Tree Algorithms. ACM Transactions on Database Systems, 18(1):51--101, 1993.

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