NTP WG INTERIM MEETING, 2019-05-29
==================================


Participants
------------

Karen O'Donoghue, Dieter Sibold, Thomas Peterson, Tal Mizrahi, Danny Mayer, Denis Reilly, Harlan Stenn, Miroslav Lichvar, Kristof Teichel, F. Gont, Watson Ladd


1.  Administrative and Agenda Bashing

	- No agenda bashing
	- No objection to record the meeting
	- Tal: Slides are not online yet
	
2.  TICTOC quick document status

	- YANG data model is published as RFC 8575
	- IEEE 1588 Enterprise Profile is submitted to the IESG. 
	- After publication of the Enterprise Profile the TICTOC WG will be closed

3.  NTP quick document status
	- BCP is in Editor Queue
	- MAC draft is in Auth 48
	- Shepard writeup for the drafts NTS for NTP und Guidelines for Defining Packet
	  Timestamps are in preparation
	- Draft YANG Data Model for NTP in Yang Doctor review
	
4.  Discussion on documents (working group and individual) that have been updated	
	
	- Interleave Mode 
		- Miroslav: updated a new version. Changes in request of the last meeting. 
		  Two paragraphs have been changes. Ready to be advanced. No questions.
		- Summary: this draft is going to the IESG
		
	- Roughtime
		- nobody of the authors are online
		
	- Correction field and Short Extension field
		- Miroslav: No changes to both documents
		- Summary (Karen): will leave it on the list and looking for the progress of the 
		  various extension field draft
	
	- Port Randomization   
	  Discussion:
		- F Gont: The usage of port 123 of NTP traffic is not necessary. 
		  Various research has found that the usage of predictable port
		  number is problematic and should be avoided. 
		- F Gont: This I-D will make NTP compliant to BCP 156.
		- Danny: Presumably this draft improves security. It does not improve
		  security
		- F Gont: An attacker has to guess what port is in use. Randomization will make
		  this more difficult
		- Danny: does not accept the arguments
		- Karen: any other comments
		- Gont: this is the same discussion as with other protocols years ago
		- Gont: the problem applies to to all transport protocols
		- Karen further comments?
		- Miroslav: this is useful and should be adopted
		- Karen: any other question
		- Kristof: should be adopted to be discussed
		- F Gont: most NTP implementation already randomize the port number. 
		- Danny: Problem with a NTP instance that is a client and a server. What do you
		  gain?
		- Harlan: the client can already use other port number
		- F Gont: It is not good practice to leave this decision to the implementation
		- Danny: there are not only four variable that identify an association but five
		- F Gont: This is ture of the application layer. I speak about the transport
		  layer. This draft improves security on the transport layer
		- Watson: we had issues because of the fixed port number 123. DDoS attacks
		- Karen (without the chair's hat): we need to have better reasoning for adopting 
		  this draft.
		- F Gont: this draft follows the work of the transport area and which is 
		  formulated as a BCP. The Transport area should be involved.
		- Karen: the argument that there is a BCP is very compelling. We should seriously
		  consider to adopt it.
		- Danny: just because there is a BCP it does not apply to any protocol. 
		- Karen: That true, but we need to consider the BCP and have to formulate that
		  it does not apply to NTP.
		- F Gont: you need to make a good argument that the BCP does not apply for NTP  
		- Summary (Karen): we need to consider the BCP and have to decide later if we
		  adopt this draft.
		
		
	- Roughtime
		- Watson: Usage of MJD for leap second issues, some other changes, ...
		- Karen: further question?
		- Discussion about the need to consider delay attacks and  bounded RTT in the 
		  draft (Tal, Watson, Kristof) 
		- Summary (Karen): Please, think about this and discuss it on the mailing list

	



5.  AOB (Any Other Business?)