W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....on replication nor on search efficiency. The authors mention that there might be negative effects on search efficiency with their approach. Substantial work on distributed data access structures has also performed in the area of distributed databases on scalable data access structures, such as [16, 17]. This work is apparently relevant, but the existing approaches apply to a different physical and application environment. Databases are distributed over a moderate number of fairly stable database servers and workstation clusters. Thus reliability is assumed to be high and replication is used ....

W. Litwin, M. Neimat, and D. A. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In VLDB, pages 342--353, 1994.

....have been receiving substantial attention in the recent years. Two classes have been studied in the literature: For indexing scalable distributed databases on workstation clusters various variants of distributed search trees and hash based access structures have been investigated (see e.g. [7]) Usually they are called scalable distributed data structures (SDDS) and are characterized by a client server architecture, a medium number of nodes, and typical by some form of global coordination, such as split coordinators or global directories. For implementing global scale resource access ....

W. Litwin, M. Neimat, D. A. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. VLDB 1994: 342-353

.... object states, so the identifier in this section refers to identifier of a persistent object state in a datastore (to avoid mismatch with CORBA object reference) The proposed structure is actually a combination of distributed shared virtual memory [1] and distributed dynamic hashing proposed in [6]. 3.2 Distributed Hashing for DSVM The shared virtual memory is segmented into pages of reasonable size. Actual locations are calculated from the values of identity using perfect hash function which yields the address of distributed shared virtual memory page where the corresponding persistent ....

....to use main memory instead of disks to hold the pages. The amount of main memory in a distributed system should anyway sufficient to keep all data in the main memory. So, instead of disk address, the hash function is used to calculate the address of a site which holds a primary copy of this page [6]. When the page is cached, the calculated pointer values may be re used if refer to a page cached into the same address space, providing extremely fast access to persistent data. In terms of address space, we assume that the virtually addressable space is huge, and distinguish the following ....

[Article contains additional citation context not shown here]

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the Twentieth International Conference on Very Large Databases, pages 342--353, Santiago, Chile, 1994.

.... nodes is a technique that has been investigated in [7] The same authors have shown that, under certain assumptions, in that with that approach balanced search trees do not exist [8] The replication of the complete search structure is an approach that underlies the RP Trees proposed in [9]. In [11] a mechanisms is proposed that leads eventually to the replication of the search structures. Scalable replication of a search tree (more precisely B Tree) is proposed in [6] dB Tree) and [12] FatBTree) With scalable replication each node stores a single leaf node of the search tree, ....

....tree, the root node the search tree is replicated to every node, and the intermediate nodes are replicated such that each node maintains a path from the root to its own leaf. No search structures: in these approaches operation messages are broadcasted to all participating nodes. e.g. with RPv [9] the data is range partitioned as in B Trees but no index exists and a multicast mechanism is used. In current P2P file sharing systems like Gnutella the P2P network is used to propogate search requests to all reachable peers. Most of these approaches assume a reliable execution environment, ....

W. Litwin, M. Neimat, D. A. Schneider RP*: A Family of Order Preserving Scalable Distributed Data Structures. VLDB 1994:342-353

....approaches adopted by these servers are difficult to design and manage, adapt poorly to changes in workloads and configuration, and have limited fault tolerance. 3.2 Distributed Storage Management Distributed, persistent data structures are an active area of study. For example, LH [97] and RP [96] provide distributed tables that can grow dynamically and asynchronously. LH RS [98] and dPi tree [99] further replicate tables for fault tolerance. DDS [71] provides a distributed, persistent, and replicated hash table, on which applications such as web servers or protocol gateways are ....

Witold Litwin, Marie-Anne Neimat, and Donovan A. Schneider. RP*: A family of order preserving scalable distributed data structures. In International Conf. on Very Large Data Bases (VLDB), pages 342--353, Santiago, Chile, 1994. 3.2

No context found.

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of Order Preserving Scalable Distributed Data Structures. HPLDTD -94-012, (Feb. 1994).

No context found.

W. Litwin, M-A. Neimat, and D. Schneider. RP* : A Family of OrderPreserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

No context found.

W. Litwin, M.-A. Neimat and D. Schneider. RP* : a Family of OrderPreserving Scalable Distributed Data Structure. In Proceedings International Conference on Very Large Data Bases (VLDB), 1994.

....the subject of major effort [U94] One aspect of this research is the design of data structures for multicomputer files. A new class of data structures, termed Scalable Distributed Data Structures (SDDSs) has been proposed in [LNS93] Several other SDDSs have been proposed since [LNS93a] D93] [LNS94], LN94] VBWY94] KLR96] Their common trait is to allow for scalable distributed files that are orders of magnitude larger and faster than the traditional ones. In particular, one can create multi GByte RAM files that reside entirely in the distributed RAM of the multicomputer, and that ....

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of Order-Preserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

....LH file. For instance for k = 4, it represents only a 25 increase. The insert time is determined basically by the CPU time to send out and receive the message, and by the transfer time. For longer records, over Kbytes, or slower nets, e.g. Ethernet, the transfer time almost entirely dominates [LNS94]. LH S typical insert time should then be only 20 25 longer than the LH insert time. For a 10 Mb s net, and 1 KB record, this leads to the insert time of about 1.25 ms [LNS94] For faster nets or shorter records, the CPU time begins to dominate. The insert time of LH S becomes then closer to ....

....time. For longer records, over Kbytes, or slower nets, e.g. Ethernet, the transfer time almost entirely dominates [LNS94] LH S typical insert time should then be only 20 25 longer than the LH insert time. For a 10 Mb s net, and 1 KB record, this leads to the insert time of about 1. 25 ms [LNS94]. For faster nets or shorter records, the CPU time begins to dominate. The insert time of LH S becomes then closer to (k 1) 2 times LH time, as the servers work in parallel, but the client basically serializes the received messages. The exact figures depends on the network speed and topology. ....

[Article contains additional citation context not shown here]

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of Order-Preserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

....in particular the RAM storage 3 . The first SDDS proposed and becoming well known, was LH [LNS93] LNS96] K98] R98] Today, several SDDSs are known. They provide hash based, ordered, or multi attribute distributed access orders of magnitude faster than to traditional files [D93] KW94] [LNS94], VBWY94] KLR96] LN96] TZK96] SDDS files may also be much larger than centralized files. It is well known that some applications require the high availability (fault tolerant) schemes, delivering data despite site failure. Many static high availability schemes are known, some for ....

....of how much the file scales up. These figures translate to access times depending on the network and CPU speeds. Experiments with the LH RAM files on a Windows NT multicomputer using 100 Mbit s Ethernet, show the key search time of 200 s [B96] close to the theoretical value of 186 s in [LNS94]. On a Gbit s network, key search times should be in general under 100 s, the CPU speed then becomes a bottleneck. These times are orders of magnitude faster than for disk based files. Note that the distributed RAM of a modern multicomputer can easily hold many GB files that traditionally had to ....

[Article contains additional citation context not shown here]

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of Order-Preserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

....selection predicate or parameters. The servers return the selected records in parallel. Several SDDSs are known. In particular, the LH schemes provide the scalable distributed linear hash partitioning, 8] 10] Likewise, the RP schemes provide the scalable distributed range partitioning, 4] [11], 18] Several prototypes have implemented selected SDDSs. The SDDS prototype that we design at CERIA is the most extensive such system, to the best of our knowledge. It runs on Wintel multicomputers and is designed for any SDDS. At present, it offers several variants of LH and RP schemes. Some ....

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of OrderPreserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

....the RAM access time were a minute for the CPU, then every disk access would make the application idle for eight days. More prosaically, a simple aggregate function over a typical 4 GByte disk file with the current I O speed of a few Mb s easily takes hours. 2 other followed, D93] KW94] [LNS94], VBWY94] KLR96] LN96] TZK96] K98] R98] They allow for hash based, ordered, or multi attribute distributed files, much faster and larger than could be the more traditional ones. They also allow for efficient parallel scans of these files. Some applications require high availability ....

Litwin, W., Neimat, M-A., Schneider, D. RP* : A Family of Order-Preserving Scalable Distributed Data Structures. 20th Intl. Conf on Very Large Data Bases (VLDB), 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. A. Schneider. Rp*: A family of order preserving scalable distributed data structures. In Proceedings of the 20th VLDB Conference, 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A family of order preserving scalable distributed data structures. In Proc. 20th Int. Conf. Very Large Data Bases (VLDB), pages 342--353, September 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. A. Schneider. RP* : A family of order preserving scalable distributed data structures. In Proc. VLDB, 1994.

No context found.

Witold Litwin, Marie-Anne Neimat, and Donovan Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.

No context found.

W. Litwin, M-A. Neimat, and D. Schneider. RP*: A Family of OrderPreserving Scalable Distributed Data Structures. In Proc. of the 20th VLDB Conference, pages 342-353, Santiago, Chile, 1994.

No context found.

Witold Litwin, Marie-Anne Neimat, and Donovan Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A family of order-preserving scalable distributed data structures. In Proceedings of the 20th Conference on Very Large Databases (VLDB), pages 342--353, Santiago, Chile, 1994.

No context found.

W. Litwin, M.-A. Neimat, and D. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In Proceedings of the 20th International Conference on Very Large Databases (VLDB 20), pages 342--353, Santiago, Chile, Sep 1994.

No context found.

W. Litwin, M. Neimat, and D. A. Schneider. RP*: A Family of Order Preserving Scalable Distributed Data Structures. In VLDB, pages 342--353, 1994.

No context found.

W. Litwin, M-A. Neimat, D. Schneider: RP*: A Family of Order-Preserving Scalable Distributed Data Structures, 20 Intl. Conf. on Very Large Databases (VLDB'94), 1994.

First 50 documents

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