S. Savage, A. Collins, E. Homan, J. Snell, and T. Anderson. The Endto -end Eects of Internet Path Selection. In Proceedings of the ACM SIGCOMM '99 Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, pages 289-299, September 1999.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....can improve service availability. Studies consistently nd that, in contrast with targets of four nines or ve nines of availability (99.99 uptime or 99.999 uptime) for important services, the Internet network layer provides only about two nines of host to host 1 connection availability [2, 3, 4]. The resulting average of about 14 minutes per day of unavailability to a typical client hinders commercial sites as well as mission critical sites. 3. Disconnected operation can signi cantly improve performance. Traditional web caching is a simple example of this strategy, and several studies ....

....if users desire to do so. 57 Chapter 5 Related work 5. 1 Disconnected operations related In the context of web services, previous studies have examined the performance bene ts of web caching [1, 6] prefetching [18, 21, 22, 31] pushing updates [6] mobile code [5, 16, 7] and overlay routing [4]. Systems implementing variations of some of these techniques have been built. File caching [5] replication [43] hoarding [13, 14] and write bu ering are standard techniques for coping with disconnection for static le services. Active channels [12] provide an interface for server directed ....

S. Savage, A. Collins, E. Homan, J. Snell, and T. Anderson. The Endto -end Eects of Internet Path Selection. In Proceedings of the ACM SIGCOMM '99 Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, pages 289-299, September 1999.

....policies. They developed an abstract model of BGP and de ned a number of conditions that can be checked to make sure that BGP routes do not oscillate. For each of these conditions, they showed that the complexity of statically checking it is either NP complete or NP hard. Recently, Savage et al. [19] found that in several cases an alternate routing path exists which is superior with respect to performance metrics like round trip time, loss rate and bandwidth to the actual path taken by packets. This is not surprising since routing path selection in several deployed routing protocols is based ....

S. Savage, A. Collins, and E. Homan. The End{to{End Eects of Internet Path Selection. In Proceedings of the ACM SIGCOMM '99, Sept. 1999.

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