USC CSCI 551 - 14_paxson99b-6up (3 pages)

Previewing page 1 of 3 page document View the full content.
View Full Document

14_paxson99b-6up



Previewing page 1 of actual document.

View the full content.
View Full Document
View Full Document

14_paxson99b-6up

92 views


Pages:
3
School:
University of Southern California
Course:
Csci 551 - Computer Communications

Unformatted text preview:

Computer Communications CSCI 551 Computer Communications CSCI 551 End to end Packet Dynamics CS551 End to End Internet Packet Dynamics How do you measure Internet performance Why do people want to know Are ISPs willing to tell you What kinds of packet dynamics are observed in the network Does there exist a typical Internet path Paxson99b Bill Cheng http merlot usc edu cs551 f12 1 Copyright William C Cheng 2 Copyright William C Cheng Computer Communications CSCI 551 Computer Communications CSCI 551 Key Ideas Methodology Measure Internet traffic active measurements 2 N paths lots of details out of TCP Previous studies Focused on a small number of paths Used unrealistic traffic pings etc Paxson s study Examined nearly 1000 paths Used TCP traffic routers designed to handle TCP as common case congestion adaptive both good and bad Was extraordinarily careful used statistically valid sampling to reduce bias looked at the wire to get most confidence adjusted for TCP implementation idiosyncrasies Evaluate dynamics pathologies out of order duplication corruption bandwidth loss delay 3 Copyright William C Cheng 4 Copyright William C Cheng Computer Communications CSCI 551 Computer Communications CSCI 551 Pathologies Reordering Impact of Reordering Reordering packets arrive at receiver in a different order than they were sent On TCP fast retransmit and recovery Which assume packet loss upon receiving dup ACKs But packets may actually have been reordered Evidence Significant non trivial occurrence 10 30 connections Strongly site dependent Most egregious instances correlated with route flutter Different packets sent along different routes Can we avoid this by Waiting before sending ACK yes about 20ms waits would have detected most reordering events Reducing the dup ACK threshold possibly to 2 But these require server and client side change bottom line current techniques work Other curious effects Router forwarding lulls i e stops forwarding as if it has gone to sleep 5 Copyright



View Full Document

Access the best Study Guides, Lecture Notes and Practice Exams

Loading Unlocking...
Login

Join to view 14_paxson99b-6up and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view 14_paxson99b-6up and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?