Transport Area (TSV) BOF: STORM (STORage Maintenance) THURSDAY, March 26, 2009, 0900-1130 Morning Session I IETF San Francisco Meetings ======================================================= BOF chair: David L. Black, EMC (black_david@emc.com) Minutes - Final Version ----------------------- --- Administrivia ---- - Note Well (projected) - Blue Sheets (circulated) - Scribes (minutes, jabber) Jabber room was initially not functioning, but was fixed shortly after BOF started - Purpose of BOF: Question #1: Should an IETF WG be formed? --- Agenda Bashing --- Agenda was bashed to add discussion of possibly related work in T11 (Fibre Channel Standards Organization). The T11 vice chair (and acting chair) was present to discuss the item. -- Draft charter: Initial presentation --- Charter presented, no text bashed. -- Discussion of possible iSCSI-related work -- - Consolidated iSCSI RFC - Whether to take iSCSI to Draft Standard status, including implementation report - SAM-4 feature addition Extensive discussion indicated clear strong interest in the SAM-4 feature addition and interest in consolidating the iSCSI RFCs into a single document that could go to Draft Standard status. The resulting plan is for two drafts: 1) An rfc3720bis draft that consolidates and prunes the current iSCSI RFCs based on current implementations. The intent is that this draft be suitable for Draft Standard RFC status. 2) A "new features" draft that adds (in a backwards compatible fashion) minor new features, starting with the new SAM-4 features (but not limited to SAM-4). This draft would be intended for Proposed Standard status, and all of its contents should be optional and negotiable (e.g., via text keys at iSCSI login). The "new features" draft would not be limited to SAM-4 features, but all additions would need to be minor. IETF procedures allow everything to be rolled into one document that could be taken to Draft Standard status after a 6 month waiting period at Proposed Standard, but the sense of the discussion was that the above two- document plan is the better course of action, and that only widely implemented features in current implementations should be considered for a possible Draft Standard RFC. The BOF chair prefers that the decision as to whether to take iSCSI to Draft Standard status be deferred, and will write charter text to allow this, but not require it. Among the reasons for this are the amount of work involved and potential interactions with other RFC documents that are currently at Proposed Standard status (e.g., RFC 3723 on security). It seems to make better sense to specify iSCSI support for all new SAM-4 features (as other SCSI transports, such as FCP [SCSI over Fibre Channel] and SAS [Serial Attach SCSI]) have done, rather than only specify selected SAM-4 features. iSCSI could then define multiple text keys to allow support for these features (or groups of them) to be separately negotiated. IETF coordination with T10 is planned to be handled informally. At least David Black (BOF chair) and Fred Knight (one of the presenters on the SAM-4 topic) are active T10 participants who can facilitate this coordination. The formal IETF mechanisms for liaison with other organizations appear unlikely to be needed, but can be used if necessary. --- FC encapsulation-related work (possible) --- - iFCP Address Translation obsolescence - FC(IP) IP Protocol Number Both of these appear to be good ideas. The BOF chair is prepared to write the Internet-Draft for the first item, and the second item is for the WG to investigate whether IP Protocol number 133 (allocated for Fibre Channel in 2000) is unused and should be returned to IANA for future reassignment. - Related T11 activity (RDDP over Ethernet) T11 is the standards organization for Fibre Channel. A proposal has been submitted to T11 for it to work on hosting the RDDP (aka iWARP) protocols on Ethernet directly via IP, see T11 document 09-141v0: http://www.t11.org/ftp/t11/admin/project_proposals/09-141v0.pdf The design proposed in that document appears to require allocation of an IP Protocol number. That can only be done by the IESG, and appears unlikely, as the proposed DCRP protocol is not intended to run over public networks like the Internet, and has neither congestion control nor security. UDP encapsulation may be a more plausible way forward, but the AD (Lars) is concerned about a track record of protocols intended for closed networks "escaping" into the broader Internet. The BOF chair (David Black) is also a member of the IETF Transport Directorate and a T11 participant - he will take these concerns to the T11 meetings next week and express a desire for significant consultation between T11 and the IETF before T11 moves this work moves forward. The responsible Transport AD (Lars Eggert) concurs with and supports this course of action. --- RDDP-related work (possible) --- - MPA startup change for MPI There has been a lot of support for this relatively contained proposed work item on the mailing list. --- iSER-related work (possible) --- - Clarifications arising from InfiniBand and other use of iSER There has been support on the mailing list for this proposed work item. --- Any other proposed work items --- Nothing additional proposed. --- Draft charter: Text bashing, round 2 --- No further bashing. --- WG formation discussion --- Clear interest in forming a WG, with a desire to limit travel. Work item discussion took a bit longer. Different people are interested in different work items. A discussion of whether any work items should be deleted from the initial list in the charter did not identify any to be deleted, but acknowledged the BOF chair's preference to not have the charter commit up front to taking iSCSI to Draft Standard status. With that change, the "rough consensus" of the meeting (supported by some comments in the jabber room and emails sent to the list) is that a WG should be formed to take on essentially the plan of work in the draft charter. Discussion about travel and meetings reflected a concern about travel in current economic conditions and a desire to conduct as much work as possible on the mailing list. Beyond that, face-to- face meetings are sometimes needed to resolve issues and make progress - in line with current IETF practice, if a meeting is needed during an IETF meeting week, it should be scheduled without significant consideration of where that IETF meeting is located. The expected default for the WG will be to not meet face-to-face unless there are clear issues that require a meeting. Interim meetings were suggested as a possibility, but there does not appear to be a geographic concentration of interested people that results in an obvious site; the BOF chair is aware of likely participants from both coasts of the US, plus China, Switzerland and Israel. Iceland was (partly in jest) suggested as approximately equally inconvenient for all concerned. --- Mailing List Usage --- Organization of the BOF was conducted on the existing ips@ietf.org and rddp@ietf.org mailing lists, with the imss@ietf.org list cc:'d. After some discussion, the mailing list plan will be: - Form new storm@ietf.org mailing list for new STORM WG. - Announce this list on the IMSS, IPS and RDDP lists, inviting people to subscribe to the new STORM list. - No further STORM-related activity on the IMSS list after this announcement (and perhaps a few reminders). - Set IPS and RDDP auto-responders to indicate existence of STORM list. - After some period of time, close down the RDDP list, ask the IETF secretariat to set up an auto-responder for email to rddp@ietf.org indicating that storm@ietf.org should be used. - Leave IPS list open with auto-responders modified as above. - Leave IMSS list as it currently is. --- Milestones --- In order to have a charter that can be approved, milestones are needed. A small amount of discussion yielded the following guidelines: - Small stuff (e.g., MPA fix) should make it to Working Group Last Call (WG LC) this year - Other larger things (e.g., SAM-4) may need until latter part of next year to reach WG LC. - iSCSI: The consolidated iSCSI draft may be a 2 year project, but there should be a decent draft in about 1 year. - The charter will allow lazy evaluation of decision about whether to take iSCSI to Draft Standard, and hence not propose any milestones for that. --- Next steps --- - BOF chair to revise charter (with milestones) and send to mailing lists for review. - Assuming no significant objections to WG creation are raised, the secretariat will be asked to create a storm@ietf.org mailing list. - The actual storm WG formation request will probably go to the Area Director (Lars) for presentation to the IESG in about a month. Attendance: 14 names on blue sheet, approximately 8 additional jabber room participants.