The Trouble with Triggers

1
The Trouble with Triggers

• Bernard Aboba
• Microsoft
• Draft-ietf-dhc-ipv4-dna-04.txt
• ALIAS BOF
• IETF 58
• Minneapolis, MN
• Tuesday, November 10, 2003
• http//www.drizzle.com/aboba/IETF58/ALIAS/

2
What is a Trigger?

• A lower layer indication
• Also known as a hint in DNA
• Whats in a word?
• A hint can be wrong
• A trigger implies action
• Thought for the day
• Implementations MUST be robust in the face of
  misleading hints

3
Bad Hint, No Donut

• Using SSID as a hint in moving from one private
  network to another
• SSID names the ESS, not the network prefix
• Result host moves from one default SSID to
  another, concludes it is on the same link. Wrong!
• Link down
• Media Sense implemented in Windows 2000
• Some applications (foolishly) consumed the hint
• Result tear down of TCP connections after
  momentary connectivity loss
• Link up
• When does link becomes usable for IP
  connectivity?
• IEEE 802.11-1999 Reassociation Response
• IEEE 802.11i Completion of 4-way handshake
• Result
• DHCP packets go out prior to completion of L2
  auth are dropped
• TCP connections send, go into slow start

4
Importance of Robustness

• Questions
• What happens if the hint is wrong?
• Can the host test the veracity of the hint?
• Can the host recover if the hint is wrong?
• Example DNA
• Reachability test required regardless
• If reachability test disagrees with hint,
  hint loses

5
Some Questions

• For purposes of ALIAS, do we know which hints are
  strong or weak?
• How does ALIAS verify hints?
• How does ALIAS recover from misleading hints?

6
Concept Strong vs. Weak Hints

• DNAv4 draft talks about Strong vs. Weak hints
• Distinction refers to probability of the hint
  being misleading
• Goal encourage creation of strong hints
• Question does host behavior change regardless of
  the strength of the hint?
• DNA No, reachability detection required anyway
• Motto Trust but Verify

7
Feedback?