P. J. Varman and R. M. Verma. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3). IEEE, May 1997. 16  Home/Search   Document Not in Database   Summary   Related Articles   Check

....extremely space efficient comprehensive versioning. In addition to the significant file system work in versioning, there has been quite a bit of work done in the database community for keeping versions of data through time. Most of this work has been done in the form of temporal data structures [2, 23, 24, 44, 45]. Our directory structure borrows from these techniques. The log structured data layout was developed for writeonce media [33] and later extended to provide write performance benefits for read write disk technology [36] Since its inception, LFS has been evaluated [3, 28, 35, 38] and used [1, 7, ....

P. J. Varman and R. M. Verma. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3). IEEE, May 1997. 16

....blocks, to store N elements; supports updates in O(log B N) I Os; and performs one dimensional range queries in optimal O(log B N T B) I Os, where T is the number of reported elements. Using a general technique by Driscoll et al. 14] persistent versions of the B tree have also been developed [11, 26]. A persistent data structure maintains a history of all updates performed on it, such that queries can be answered on any of the previous versions of the structure, while updates can only be performed on the most recent version (thus creating a new version) A persistent B tree uses O(N B) ....

....number of updates performed, and updates and range queries can be performed in O(log B N) and O(log B N T B) I Os, The type of persistence we describe here is often called partially persistent as opposed to full persistence where updates can be performed on any previous version. respectively [11, 26]; note that the structure requires that all elements stored in it during its entire lifespan are comparable, that is, that the elements are totally ordered. In the RAM model, several linear space planar point location structures that can answer a query in optimal O(log 2 N) time have been ....

[Article contains additional citation context not shown here]

P. J. Varman and R. M. Verma. An e#cient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391--409, 1997.

....CVFS to provide efficient comprehensive versioning. In addition to the significant file system work in versioning, there has been quite a bit of work done in the database community for keeping versions of data through time. Most of this work has been done in the form of temporal data structures [2, 22, 23, 44, 45]. Our directory structure borrows from these techniques. The log structured data layout was developed for write once media [33] and later extended to provide write performance benefits for read write disk technology [36] Since its inception, LFS has been evaluated [3, 27, 35, 38] and used [1, ....

Peter J. Varman and Rakesh M. Verma. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering,

....the operations in temporal databases [ST99] The techniques of Driscoll et al. motivated a number of techniques for external persistent versions of the B tree. Lanka and Mays [LM91] proposed the first, fully persistent B tree. Subsequent works by Becker et al. BGO 96] and Varman and Verma [VV97] refined the approach, by addressing a number of practical issues, such as thrashing, and retirement of old versions to tertiary storage. 2.2 Multidimensional range search in databases To a large extent, the initial development of multidimensional range search techniques for databases took ....

P. J. Varman and R. M. Verman. An e#cient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391--409, 1997.

.... approach is to exploit the monotonicity of the temporal dimension, and transform a 2 dimensional spatial access method to become partially persistent [38, 21, 14, 16, 32] A partially persistent structure logically stores all its past states and allows updates only to its most current state [9, 20, 4, 39, 18, 29]. A historical query about time t is directed to the state the structure had at time t. Hence, answering such a query is proportional to the number of alive objects the structure contains at time t. That is, it behaves as if an ephemeral structure was present for time t, indexing the alive ....

....maintained for each time instant. Since consecutive trees do not differ much, common (overlapping) branches are shared between the trees. While easy to implement, overlapping creates a logarithmic overhead on the index storage requirements [29] Conceptually, the multi version approach [20, 4, 39, 18, 32] also maintains a 2 dimensional index per time instant, but the overall storage used is linear to the number of changes in the evolution. In the rest we use a partially persistent R Tree (PPR Tree) 18, 32] a short description of a PPR Tree appears in the next section) Our approach for ....

[Article contains additional citation context not shown here]

P.J. Varman and R.M. Verma. An Efficient Multiversion Access Structure. IEEE Transactions on Knowledge and Data Engineering, Vol. 9, No 3., pages 391--409, 1997.

....problems. Using standard persistent techniques [137, 70] a persistent B tree can be designed such that updates can be performed in O(log B N) I Os and such that any version of the tree can be queried in O(log B N T=B) I Os. Here N is the number of updates and the tree uses O(N=B) disk blocks [36, 147]. In string applications a data element (string of characters) can often be arbitrarily long or different elements can be of different length. Such elements cannot be manipulated efficiently in standard B trees, which assumes that elements (and thus routing elements) are of unit size. Ferragina ....

P. J. Varman and R. M. Verma. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391--409, 1997.

....[9] An element can be inserted deleted in O(log B N) I Os. A standard B tree answers queries only on set of elements currently in the structure. A persistent (or multiversion) B tree, on the other hand, supports range queries in all states (versions) of the data structure in O(log B N K=B) I Os [8, 18]. Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2 ( 8, 18] A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N=B) disk blocks and ....

....on the other hand, supports range queries in all states (versions) of the data structure in O(log B N K=B) I Os [8, 18] Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2 ([8, 18]) A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N=B) disk blocks and supports range queries at any time in O(log B N K=B) I Os. 3 Data Structure for Moving Points in R In this section we consider queries of type Q1. If we interpret time as the ....

P. J. Varman and R. M. Verma, An efficient multiversion access structure, IEEE Transactions on Knowledge and Data Engineering, 9 (1997), 391--409.

....[9] An element can be inserted deleted in O(log B N) I Os. A standard B tree answers queries only on set of elements currently in the structure. A persistent (or multiversion) B tree, on the other hand, supports range queries in all states (versions) of the data structure in O(log B N K B) I Os [8, 18]. Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2 ( 8, 18] A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N B) disk blocks and ....

....on the other hand, supports range queries in all states (versions) of the data structure in O(log B N K B) I Os [8, 18] Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2 ([8, 18]) A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N B) disk blocks and supports range queries at any time in O(log B N K B) I Os. 3 Data Structure for Moving Points in R In this section we consider queries of type Q1. If we interpret time as the ....

P. J. Varman and R. M. Verma, An e#cient multiversion access structure, IEEE Transactions on Knowledge and Data Engineering, 9 (1997), 391--409.

....inserted deleted in O(log B N) I Os. Note that a standard B tree, also called an ephermal structure, answers queries only on the current set of elements. A persistent (or multiversion) B tree, on the other hand, supports range queries in all states (versions) of the index in O(log B N K=B) I Os [9, 32]. Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2.2 ( 9, 32] A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N=B) disk blocks and ....

....B tree, on the other hand, supports range queries in all states (versions) of the index in O(log B N K=B) I Os [9, 32] Updates can be performed in O(log B N) I Os on the current structure, and we refer to the structure existing after T updates as the structure existing at time T . Lemma 2. 2 ([9, 32]) A persistent B tree constructed by performing N updates using O(log B N) I Os each, uses O(N=B) disk blocks and supports range queries at any time in O(log B N K=B) I Os. 3 An Index for Moving Points in R In this section we consider queries of the type Q1. We assume the points in S move along ....

[Article contains additional citation context not shown here]

P. J. Varman and R. M. Verman. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391--409, 1997.

....the Interval Tree [Ede83] which is an ephemeral memory based structure with good worst case performance into the BIT, which is disk based, partially persistent and well paginated. The BRT makes an R tree [B 90] partially persistent, following the approach of the MVB [B 93] and MVAS [VV97] Like those structures, the BRT is a directed acyclic graph of pages. The structure is then formed by several logical R trees, representing the evolution of objects in the transaction time sense. The 2R tree uses two R trees (named front and back R trees) to index bitemporal data. The bitemporal ....

P.J. Varman and R.M. Verma. An efficient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391--409, 1997.

No context found.

P. J. Varman and R. M. Verma. An ecient multiversion access structure. IEEE Transactions on Knowledge and Data Engineering, 9(3):391-409, 1997.

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