Jacobson, V., "Congestion Control and Avoidance," Proc. ACM SIGCOMM `88, Stanford, CA, August 1988. The paper describing improvements to TCP to allow use over Wide Area Networks and through gateways connecting networks of varying capacity. This work was a starting point for the NFS Dynamic Retransmission work.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... has been in a large part due to the existence and wide scale use of a standardized, reliable unicast transport protocol, the Transport Control Protocol (TCP) RFC793] In addition to providing for reliable data transport, TCP also provides effective, end to end congestion control mechanisms [Jacobson88, Stevens97]. At present, multicast transport solutions lack such standard, proliferated approaches to congestion control that operate robustly across a wide range of network scenarios and support fairness when deployed alongside existing network transport standards . Best effort (i.e. unreliable) native ....

V. Jacobson, "Congestion Control and Avoidance", Proc. of SIGCOMM 1988, pp. 314-328.

....almost as well as as JSQ prefetching: they allows for nearly 100 link utilization, immediate commencement of playback and instanta3 neous response to viewer interactions. Our decentralized prefetching protocol employs window flow control; it is inspired by the Transmission Control Protocol (TCP) [7, 1] widely used in the Internet. For simplicity, assume that each server is responsible for exactly one connection. Admission control ensures that all link utilizations do not exceed 95 . Our basic decentralized prefetching protocol works roughly as follows. The server maintains a send window, ....

V. Jacobson. Congestion control and avoidance. In Proceedings of SIGCOMM '88 Symposium, August 1988.

....rate: one segment per round trip time. Fast retransmit and fast recovery are used together to detect and repair losses of data, Fig. 1. Plot of Instantaneous Bandwidth of a Local and a Remote Stream Contending for Bandwidth after which transmission continues normally with a reduced TCP window [2] [4]. A typical plot of two streams contending for bandwidth is shown in Figure 1. The bandwidth used increases rapidly during the SS phase. Stream 1 then loses some data as the bandwidth capacity is reached but then quickly recovers using the FR algorithm and continues with SS. Finally both streams ....

V. Jacobson, M.J. Karels, Congestion Control and Avoidance, Proc. ACM SigComm 1988, 314-329, November 1988. (http://citeseer.nj.nec.com/).

.... Internet Protocol (IP) technology worldwide has been in a large part due to the existence and wide scale use of a standardized, reliable unicast transport protocol (i.e. TCP) In addition to reliable data transport mechanisms, TCP also provides effective, end to end congestion control mechanisms [Jacobson88, Stevens97]. At present, multicast technology lacks such best practice approaches to congestion control that operate robustly across a range of network scenarios and support fairness when deployed alongside existing network transport standards. MOTIVATION AND ISSUES Best effort (i.e. unreliable) ....

V. Jacobson, "Congestion Control and Avoidance", Proc. of SIGCOMM 1988, pp. 314-328.

....1 person month client (excluding asynchronous writes) 2 person months client asynchronous writes 1 person month prove asynchronous WRITE performance, especially for small files. Adaptive retransmission strategies to improve the behavior of NFS over UDP (as described in [Nowicki89] derived from [Jacobson88]) and the use of TCP to improve performance over wide area networks [Macklem91] are applicable to NFS Version 3. NFS Version 3 relaxes the 8KB limitation on the data portion of a READ or WRITE request, permitting more efficient use of TCP. Three efforts to revise the NFS protocol are related to ....

Jacobson, V., "Congestion Control and Avoidance," Proc. ACM SIGCOMM `88, Stanford, CA, August 1988. Describes TCP performance improvements over WANs and gateways. This work was a starting point for the NFS Dynamic Retransmission work.

....at the destination s discretion. These signal streams may replace previously selected streams. In these instances, the previously selected stream must be unreserved with an unreservation packet. 24 Alternatively, instead of relying on a single sample, a fast recursive algorithm such as that in [33] may be used to dynamically estimate the expected packet round trip time based on a sequence of received packets. Reservation Based Multicast 30 3.2.3 Shared reservations for multiple destinations The preceding describes routing from a set of RPs to each destination individually, without regard ....

V. Jacobson, "Congestion Control and Avoidance," Proc. ACM SIGCOMM '88, pp. 314329, Aug. 1988.

....traffic scenarios which will yield improved congestion control on a large internetwork. 1 Introduction The explosive growth in computer networks over the past decade has brought with it the problem of congestion. Much research done since then looks at ways of dealing with this problem [Jai86] [JK88]. MK92] WC91] The work has primarily encompassed the design, modification and extension of protocols and routers and detailed analysis of traffic characteristics on networks. An offshoot of this research has been the development of powerful simulation tools [Kes88] MF] BP96] to help analyze ....

....strategy of TCP is to send packets into the network without reservation and then react to observable events which occur. Early work in TCP congestion control include Nagle s algorithm [Nag84] and Clark s strategy for avoiding the Silly Window Syndrome [Cla82] A classic paper by Van Jacobson [JK88] proposed several TCP extensions to specifically address the congestion control problem. These include slow start, the Jacobson Karels algorithm for estimating RTT and a congestion avoidance algorithm. Karn and Partridge proposed two small but significant changes to alogrithms used in the ....

V. Jacobson and M. Karels. Congestion Control and Avoidance. In Proceedings of SIGCOMM '88, 1988.

....Adaptive retransmission Most client implementations use either an exponential back off strategy to some maximum retransmission value, or a more adaptive strategy that attempts congestion avoidance. Congestion avoidance schemes in NFS request retransmission are modelled on the work presented in [Jacobson]. Nowicki] and [Macklem] describe congestion avoidance schemes for NFS over UDP. Caching policies The NFS version 3 protocol does not define a policy for caching on the client or server. In particular, there is no support for strict cache consistency between a client and server, nor between ....

Jacobson, V., "Congestion Control and Avoidance," Proc. ACM SIGCOMM `88, Stanford, CA, August 1988. The paper describing improvements to TCP to allow use over Wide Area Networks and through gateways connecting networks of varying capacity. This work was a starting point for the NFS Dynamic Retransmission work.

No context found.

Jacobson, V., "Congestion Control and Avoidance," Proc. ACM SIGCOMM `88, Stanford, CA, August 1988. The paper describing improvements to TCP to allow use over Wide Area Networks and through gateways connecting networks of varying capacity. This work was a starting point for the NFS Dynamic Retransmission work.

No context found.

V. Jacobson. "Congestion Control and Avoidance". In Proc. ACM SIGCOMM, August 1988.

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