Web100 Status and Impact

Slides: Web100
Status and
Impact

[Start Here (html)] [Slide Sorter (html)]

Download:



Page 1: Web100

Page 2: Pushing up performance

Page 3: The Wizard Gap

Page 4: The Wizard Gap

Page 5: TCP "tuning" requires experts

Page 6: TCP tuning is painful debugging

Page 7: Web100

Page 8: When there is a problem, just ask TCP

Page 9: TCP triage instrument

Page 10: A Demo

Page 11: Diagnosing path bottlenecks

Page 12: Diagnosing path bugs

Page 13: Localizing path problems

Page 14: Sender and receiver bottlenecks

Page 15: Web100 Project Overview

Page 16: Current Web100 status

Page 17: Web100 Summary

Page 18: end of part 1

Page 19: The Other Problem:

Page 20: The TCP model

Page 21: Fitting the model 1

Page 22: Fitting the model 2

Page 23: Fitting the model 3

Page 24: Fitting the model 4

Page 25: The other problem is the loss rate

Page 26: end of part 2

Page 27: Implications to the network

Page 28: Danger - new load levels

Page 29: Why not before Web100?

Page 30: We expect to see line rate traffic

Page 31: Problems

Page 32: No deployed queue management

Page 33: Demo 2

Page 34: No deployed QoS

Page 35: Pervasive broken link layers

Page 36: No models for traffic sharing

Page 37: Shared gigabit problem

Page 38: No SLA quantification

Page 39: New load levels

Page 40: The common theme

Page 41: end of part 3

Page 42: Implications to users

Page 43: How well are users doing today?

Page 44: We need independent measures

Page 45: The General Plan

Page 46: Passive measurements

Page 47: Window size measurements

Page 48: Observed window sizes

Page 49: Observed window sizes

Page 50: Receiver window sizes

Page 51: Receiver window sizes

Page 52: end of part 4

Generated by MagicPoint