RE: Notes from IPS meeting

[Eddy Quicksall <eddy_quicksall@ivivity.com>]

Where can I find the "SCSI Command Ordering Draft"?

Eddy

-----Original Message-----
From: Elizabeth G. Rodriguez [mailto:Elizabeth.G.Rodriguez@123mail.net] 
Sent: Friday, March 28, 2003 12:19 AM
To: ips@ece.cmu.edu
Subject: FW: Notes from IPS meeting

Here are the minutes from the IPS meeting last week.
Please review and let David and/or I know of any
comments/corrections/issues by 5pm EST on Tuesday April 1.

Thanks,

Elizabeth Rodriguez
----


IPS Working Group - Tuesday, March 18, 2003

Status of Drafts -

Per the IETF ID Tracker, located at
https://datatracker.ietf.org/public/pidtracker.cgi

Most thru working group last call.
A few approved for publication.

Status:  Publication Requested (these are standards track documents
which have just recently been forwarded to the ADs for initial review)
	Auth MIB
	FC Mgmt MIB
	iFCP MIB
	iSCSI MIB

Status: AD Evaluation
	FCIP SLP (has completed expert review.  Same expert review
applies to iSCSI SLP)
	iSCSI SLP
	Stringprep
	iSNS

Status: IESG Evaluation/AD Followup
	Naming and discovery - Back to ADs for review after revisions
made in response to IESG feedback

Status: IESG Evaluation/Revised ID needed
	Boot draft sent back due to security problems.  Issues discussed
later in session.

Status: RFC Editor Queue (Approved for Publication)
FC Common Encapsulation
IPS Security
iSCSI
FCIP
iFCP
These documents are in the RFC Editor Queue and are pending references.
To see the dependencies these drafts are awaiting, see
http://www.rfc-editor.org/queue.html.

Long poll on the iSCSI RFC assignment is waiting on the AES Counter
draft. Possible expedition after references are completed.

Question was asked about status of FC-BB-2 publication.  Craig Carlson
(T11.3 Chair) stated that the ANSI publication process is slow and final
publication may not  be available for 6 months.

Status: ID Exists
	Three of these drafts (SCSI MIB, FCIP MIB, iSNS MIB) are almost
ready to be forwarded.  Nit checks have already occurred, and the drafts
are being reviewed by Keith McCloghrie.

	Three of these drafts (fcencaps-wglc, fcip-wglc, ifcp-wglcc) are
drafts written in response to WG Last Call comments, and these drafts
will soon be expired.

	The Name-Ext draft is a new draft approved as a WG item in
Atlanta, and discussed later in the session.

---

Boot draft
Presented by John Hufferd, on behalf of Prasenjit Sarkar

The boot draft has been sent back to the working group for further work,
due to security issues.
Need to address boot communications security only; Boot image integrity
outside purview
Software stage: not allowed in a secure boot because of integrity issues

The boot environment is many times extremely constrained verses a fully
operational environment.  A limit of 64K is possible.  Need to keep
security fairly straightforward; getting too complex takes things
backwards. 


Proposal:  Use IPsec (common in wireless networks) takes care of IP addr
initialization problem
	Bernard Aboba suggest use DHCP auth like ESP in IPsec
	Question for the IPS WG do we want another protocol at boot
	How to use IPsec w/o IP address - taken care of in wireless
community.

Bernard Aboba: 
Resources in boot processes very constrained.
IPsec in boot PROM is possible, but IKE definitely will not fit.
DHCP-AUTH probably works, but IPsec with IKE probably doesn't.
Need to consider a staged approach. IPsec is reasonable at stage two.

Recommendation from presentation:
Boot stage:  Use IPsec
Not allowed:
Pre-shared keys (dynamic IP addresses)
Aboba suggest handling like iSCSI (aggressive mode)
	Accepted
	Public keys (disallowed by IPS Security draft)
	Manual Keying (provisioning hassle)

Comments:
If someone is really interested in security, they want it all the way
down to level zero boot.
How do we integrate security mechanisms in the boot loader?

Not every mode of operation needs to be secure, but at least one mode
does need to be secure.  
Verification of boot image is very difficult.  

DHCP-Auth makes sense, at boot, security depends on site security usage.
If the second stage boot requires IPSEC, then the adapter better have
that ability available. Need to be able to 'tickle' the IPsec
functionality on the hardware.  

Big concern is that the Key probably needs to be in flash. Key exchange
may not happen - instead may use manual key for only the boot, and the
SA is torn down immediately after booting. There is a basic provisioning
problem here.
  
Direction for draft -- Two phase process.
	First:  DHCP with DHCP-auth as necessary. 
	Second: iSCSI to get boot image.  IPsec used here, with simple
key management.
		  Pre-shared keys are better than manual keys, but
provisioning problem 
		  is the same for both if they are burned into the card.

		  Pre-shared allows a session key to be generated so
lifetime is longer.

Comment by Kevin Gibbons:  iSNS has feature where boot target can be
specified.  This is as a addition to SLP.  

---
iSCSI naming extension draft
Presented by Mallikajurn

The T10 had two votes, the Grand Unified SCSI device name format passed
first vote in working group, but was narrowly defeated in 2nd vote at
plenary.  This probably implies that this is not dead yet, but notion of
common format with iSCSI is dead.  Would have another form.  

This draft does not defined LU name, but Rob Elliot's proposal requires
transports to defined the syntax of a logical unit name.

Direction for draft: Keep it alive here.  Add Rob Elliot's additional
requirements in the T10 proposal which requires transports to define
syntax of Logical Unit Name

---

SCSI Command Ordering Draft 
Presented by Mallikajurn

Motion to accept this as an official working group item.  No objections.

Will accept as working group draft.  
This is an implementer's notes draft.  We can not produce too many such
drafts.
In Vienna will continue discussion and decide if other implementation
observations (if any) need to be rolled in, or if this can proceed stand
alone.