Sujata Banerjee, Victor O. K. Li, and Chihping Wang. Distributed database systems in high-speed wide-area networks. IEEE Journal on Selected Areas in Communications, 11(4):617--630, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....this text is included. der of gigabits second is becoming available. The revolution in broadband networks aects the de sign of database management systems. The most important issue in improving the performance of a network wide database system is how to use the network bandwidth effectively[Banerjee et al. 1993, Banerjee et al. 1998, Nishio et al. 1996] and this is contrary to the conventional systems where the mini mization of data volume transmitted in (narrowband) networks had been considered as the primary factor in performance improvement. Here, the question is how we can make efficient use of ....

Banerjee, S., Li, V.O.K., and Wang, C. (1993). Distributed Database Systems in High-Speed Wide-Area Networks. IrEEE J. Selected Areas in Comm., 11(4):617-630.

....communication environments [3] In an asymmetric environment, broadcasting the information is an effective way of making the information available simultaneously to a large number of users. For asymmetric environment, several researchers have proposed algorithms for designing broadcast schedules [1, 2, 3, 5, 6, 7, 8, 14, 15, 16, 19, 18, 20, 22, 24, 25, 24, 26, 27, 36, 38, 39]. We consider a database that is divided into information items. The server periodically broadcasts these items to all clients. A broadcast schedule determines when each item is transmitted by the server. We present a new approach to design broadcast schedules that attempts to minimize the ....

S. Banerjee, V. O. K. Lee, and C. Wang, "Distributed database systems in high-speed widearea networks," IEEE Selected Areas in Communications, vol. 11, pp. 617--630, May 1993.

....item that it needs. It is important to minimize the access time so as to decrease the idle time at the client. Several researchers have considered the problem of minimizing the access time [1, 4, 7, 9, 14, 15, 20, 21] Note that, in general, a client may request multiple items simultaneously [6, 5, 8, 12]. In this case, the access time may depend on the number of items requested. Also, the client would expect to receive mutually consistent versions of the requested items. In this paper, similar to some of the past work (e.g. 1, 7, 9, 20] we consider the simplest case where a client only ....

....[12, 8] and teletext and videotex [3, 4, 21, 11] systems. The problem of data broadcasting has received renewed attention lately. The existing schemes can be roughly divided into two categories (some schemes may actually belong to both categories) Schemes attempting to reduce the access time [4, 1, 6, 5, 8, 15, 9, 21] and schemes attempting to reduce the tuning time (or power consumption) 10, 13, 14] Ammar and Wong [4, 21] have performed extensive research on broadcast scheduling and obtained many interesting results. Our square root rule is a generalization of that obtained by Ammar and Wong. Algorithm A ....

S. Banerjee, V. O. K. Lee, and C. Wang, "Distributed database systems in high-speed wide-area networks," IEEE Selected Areas in Communications, vol. 11, pp. 617--630, May 1993.

....scheme would work well, the update mechanism (described later) would prove to be a severe limitation. We have concentrated our efforts in developing a new concurrency control protocol called Send on Demand that works better than traditional protocols (example, locking) in a high speed environment [10]. In this paper, we present a very detailed comparison study between send on demand and locking at different data rates and different 2 read access distributions 1 . Then going a step further, we propose and compare two hybrid CC schemes, combining the datacycle concept with the send on demand ....

....have been obtained. However, in this paper, those issues are not addressed in order to keep the paper focussed on the concurrency control problem. For the sake of continuity, the next section deals with our previously reported research results, albeit very briefly. Details may be obtained in [10]. 2 Previous Research DDBSs in a high speed environment have some unique characteristics. These were identified in [10] based on which the new concurrency control protocol termed send on demand was developed. The most important criterion in a high speed wide area network is the communication ....

[Article contains additional citation context not shown here]

S. Banerjee, V. O. K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks," IEEE Journal on Selected Areas in Communications, Special Issue on Gigabit Network Protocols and Applications, vol. 11, pp. 617--630, May 1993.

....data items. This can also be viewed as a scheme where the primary copy of each data item is stored at the DB server, and a floating copy of the data item migrates from client to client under the supervision of the server. This scheme has a similar flavor to the send on demand scheme proposed in [6]. However, in that scheme, the data items were migrated from site to site according to the demand generated. With the location change, the ownership of the data item is transferred to the new site, thus making the recovery mechanism difficult (owing to the distribution of the log records) The ....

S. Banerjee, V.O.K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks, " IEEE Journal on Selected Areas in Communications, vol. 11, pp. 617--630, 1993.

....systems. In this paper, a special case of distributed systems, that of distributed database systems is discussed in the above context. We refer to these as gigabit networked databases (GNDB) Traditional data access and data sharing techniques are not expected to scale to gigabit network rates [6, 11, 10, 14, 2]. Thus if any advantages of a high speed network are to be realized, new schemes are required, that can efficiently utilize the huge bandwidths available. Towards this end, assuming a client server distributed database system in a shared nothing environment, in [1] we proposed a lock based ....

....data migration or data shipping in a sharednothing environment. The recovery scheme described in [8] supported data shipping in a shared disk environment. Note that the recovery scheme in shared nothing systems that support data migration is quite difficult, as supported by previous work in [13, 2]. The scheme described here not only achieves the objective of failure resiliency, but is also efficient in the number of message passing rounds, as warranted in a high speed network environment. We do not currently discuss granularity issues, and assume that a data item is the unit of concurrency ....

S. Banerjee, V. O. K. Li, and C. Wang. Distributed Database Systems in High-Speed Wide-Area Networks. IEEE Journal on Selected Areas in Comm., 11(4):617-- 630, 1993.

....read oriented transaction throughputs [1 4] is a unique architecture that attempts to achieve the above goal. The basic architecture of the datacycle database machine is centralized. However, a distributed version of the architecture was proposed in [2] although not covered in much detail. In [5], the datacycle concept was applied to a wide area distributed system, and a simple performance model was developed 1 . It was found that the datacycle scheme worked extremely well in hiding the communication latency incurred in wide area networks. The latency problem in wide area environments ....

....called the storage pump) periodically broadcasts the entire database on a high speed broadcast bus. User access units (UAU) equipped with hardware data filters extract 1 Although the distribution of data is of utmost importance in a distributed database system, a special case was studied in [5] where the data was centrally stored, but the database sites were distributed. FIBER OPTIC BUS UAU UAU STORAGE PUMP UAU: User Access Unit UM : Update Manager UAU UAU UAU UM Upstream Network Figure 1: A schematic diagram of the datacycle architecture relevant parts of the database on the fly ....

[Article contains additional citation context not shown here]

S. Banerjee, V. O. K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks," IEEE Journal on Selected Areas in Communications, Special Issue on Gigabit Network Protocols and Applications, vol. 11, pp. 617--630, May 1993.

....read oriented transaction throughputs [1 4] is a unique architecture that attempts to achieve the above goal. The basic architecture of the datacycle database machine is centralized. However, a distributed version of the architecture was proposed in [2] although not covered in much detail. In [5], the datacycle concept was applied to a wide area distributed system, and a simple performance model was developed. It was found that the datacycle scheme worked extremely well in hiding the communication latency incurred in wide area networks. The latency problem in wide area environments is of ....

....architecture overcomes the I O bottleneck. The broadcast mechanism does not have to be a physical broadcast bus. The choice of the physical broadcast mechanism does not affect the discussion here, and will not be commented on. 3 Datacycle Performance A simple performance study was done in [5, 8], and some of the results presented in this paper are derived from that research. The performance of the datacycle architecture depends on the fraction of queries generated in the system. The response time of queries with large access sets or whose access sets cannot be pre determined ....

[Article contains additional citation context not shown here]

S. Banerjee, V. O. K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks," IEEE Journal on Selected Areas in Communications, Special Issue on Gigabit Network Protocols and Applications, vol. 11, pp. 617--630, May 1993.

....is, and unless newer protocols are developed that are distance independent, scalable performance will not be achieved. 3 Latency Reduction Techniques Recently, the problem of speed of light latency reduction has started receiving some attention in the context of distributed database systems [5, 6, 7, 8], the world wide web (WWW) 2, 9, 10] and the file transfer protocol (ftp) 11, 12] All of the above schemes attempt to localize data close (by caching or data migration) to the user in order to reduce round trip network interactions. Further, most of the above schemes take into consideration ....

....The problem with this scheme is that it does not get rid of the latency for subsequent document accesses. Similarly, the authors of [10] advocate the development of a function similar to GETLIST called MGET to reduce round trip interactions beyond the TCP connection set up. The schemes in [5, 6, 7, 8] perform dynamic data migration from client to client in order to substantially reduce the average response time per distributed database transaction. In [6, 7] the send on demand concurrency control scheme was proposed in which each transaction broadcast its data access requirements to all the ....

[Article contains additional citation context not shown here]

S. Banerjee, V. O. K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks," IEEE Journal on Selected Areas in Communications, vol. 11, pp. 617--630, May 1993.

....done is to hide the latency by designing innovative protocols. Relatively little work has been done in the area of designing new protocols that specifically address the communication latency problem. The send ondemand concurrency control scheme was one such algorithm that addressed this problem [3 5]. In [3] the datacycle scheme, conceived of in Bellcore [6 10] was studied as a solution to concurrency control in high speed networks for read only systems. The datacycle scheme hides communication latency extremely well in read only systems. However, when updates are present, its performance ....

....hide the latency by designing innovative protocols. Relatively little work has been done in the area of designing new protocols that specifically address the communication latency problem. The send ondemand concurrency control scheme was one such algorithm that addressed this problem [3 5] In [3], the datacycle scheme, conceived of in Bellcore [6 10] was studied as a solution to concurrency control in high speed networks for read only systems. The datacycle scheme hides communication latency extremely well in read only systems. However, when updates are present, its performance is ....

[Article contains additional citation context not shown here]

S. Banerjee, V. O. K. Li, and C. Wang, "Distributed Database Systems in High-Speed Wide-Area Networks," IEEE Journal on Selected Areas in Communications, vol. 11, pp. 617--630, May 1993.

No context found.

Sujata Banerjee, Victor O. K. Li, and Chihping Wang. Distributed database systems in high-speed wide-area networks. IEEE Journal on Selected Areas in Communications, 11(4):617--630, 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