Unconfirmed Minutes of the HSSG meeting in La Jolla, CA 3/10/96 Meeting brought to order by Howard Frazier at 7:00pm. 55 people were in attendance. The stated that the agenda for the meeting was: Review the PAR, Objectives, and 5 Criteria in preparation for the plenary. Mike McConnell volunteered for secretary for meeting. A foil of the Objectives for the HSSG were presented by Howard Frazier. Howard indicated that the items were not listed by priority but rather goals of the work proposed. Each item was reviewed in turn by Howard. Item number 8 was re-worded from the prior presentation to enhance the clarity of the intent. A question was asked. “Why are we supporting half duplex?.” Howard Frazier responded that it was felt that is was worth doing and didn’t add significant effort. Item #12 which poses the question about network diameter was discussed and included a discussion on the history and reasons that lead up to it’s inclusion. Bernard Daines pointed out that a recent article in Data Communication magazine erroneously stated that the HSSG is planning to enlarge the packet size. Howard emphasized that enlargement of packet size is specifically not an objective of this effort. Howard Frazier distributed copies of the January meeting minutes which contained the EMail reflectors. Howard showed foils identifying stds-802-3-hssg@mail.ieee.org and howard.frazier@sun.com as valid EMail addresses for contact. Howard stated that we need to get everyone in the 802 community aware of what this group is doing. Howard Frazier and Howie Johnson will need to go to each dot group and present and collect questions. We will need to respond to these questions by Wednesday night. The 5 Criteria was presented next. Howard Frazier stated that all project in 802 must address and answer these 5 questions. This has been distributed on EMail for review. 1) Broad Market Potential Howard Frazier changed the number of EMail participants from 165 to greater than 210. Next discussion on the xxx and yyy place holders in the description. Discussion followed on why we don’t ask who is going to implement this and use those numbers. This lead to a discussion on balanced cost. Howard took a survey of how many plan to participant in the standardization effort for the xxx and yyy values. xxx was counted as 39 and yyy was 28. 2) Compatibility with IEEE Geoff Thompson offered comments on what portion of the application space will be addressed by CSMA/CD, specifically half duplex and relationship of network diameter. Howard Frazier responded that issues such as changing some parameters will be decided within the scope of the work. 3)Distinct Identity Jim Carlo ask the reason for calling out STP and not UTP. Howard Johnson responded that Fibre Channel has several phys such as coax and fiber. Geoff Thompson suggested that a better choice would be to re-word with “balanced pair cable” rather than STP. Glenn Connery stated that a broad base of expertise in the supplier and customers doesn’t state forcefully enough that people have been doing more than just speeding things up to see if it works. Discussion followed on how to re-state the size of the 10/100 CSMA/CD market to convey how important this work is to the users. This data was inserted into the foils. 4)Technical Feasibility A correction was made to change 1.25Gb/s to 1.25gbaud by Colin Mick. Hon Wah Chin suggested changing wording which resulted in striking “be requested” from the text. Glenn Connery asked if 1000 or 800 has been fixed. Howard Frazier stated that 1000 was selected. Geoff Thompson stated that the second paragraph was a little weak. Howard Johnson suggested some additional wording around the demonstrated feasibility statement. Glenn Connery stated that reasonable testing hasn’t been addressed by this wording. Howard Frazier took the action item to add some additional wording to the paragraph for the next meeting. Paul Nicholich asked do we want to say anything about EMC or our ability to meet it. Geoff Thompson indicated that anyone selling equipment intends to meet those requirements and thus they intend to comply. Sheto VanDorn stated that ANSI temperature and humidity and environmental items are not included in the ANSI documents. Geoff Thompson said that those requirements also be met it just requires a careful design. Paul Nicholich added a statement that vendors are meeting this EMC requirement and thus it is technically feasible to expect this to comply too. Howard Frazier suggest some initial wording which was followed by discussion on minor wording changes. The finial text was “Vendors of full speed fibre channel components and systems are building reliable products which operate at 1.06gbaud and which meet worldwide regulatory and operational requirements.” 5) Economic Feasibility General discussion began on short haul vs long haul. Geoff Thompson suggested “Cost for coaxial based short run copper links are well established for full speed fibre channel.” “The cost model for horizontal copper has not yet been established.” Howard Frazier agreed to add this to the 5 criteria. Next the PAR form was reviewed. The expected vote date will be Thursday and the group formed will be 802.3z. Specifically noted: Target competition of December 98. Howard Johnson suggested a modification of the purpose to add “who” is targeted to, i.e. the existing CSMA/CS customer base. Geoff Thompson suggested that any changes would make it worse not better. Howard Frazier discussed the patent issues indicating that he has received a letter from IBM stating they will grant licenses on a non-discriminating basis and that would satisfy the IEEE requirements on the matter. Howard Frazier reviewed the entire PAR form and took that action item to fix the necessary wording changes. There was discussion with minor changes to wording made. Jim Bunder suggested a change to item number seven to replace “existing standard” and add “while maintaining maximum compatibility with the installed base of 10/100 CSMA/CD nodes, research and development, and market understanding.” Geoff Thompson replaced “market understanding” with “,network operation and management”. Howard Frazier will edit the document for presentation during the week. Howard Frazier opened the floor for anything anyone else would like to address. There was no additional comment. Howard then presented the technical presentation schedule for the coming meeting. The meeting was adjourned at 9:00pm Unconfirmed Minutes of the HSSG meeting in La Jolla, CA 3/12/96 & 3/13/96 Meeting brought to order by Howard Frazier at 1:45pm. Mike McConnell volunteered for secretary for meeting Howard re-iterated the meeting scheduled for this afternoon and the next morning. Today would begin the technical presentations for the HSSG. Del Hanson (HP) gave a presentation titled IEEE 802 Gb/s NMF Links: 1.063gbd vs. 1.250 Gbd Line Rate Issues. Major issues addressed were the areas of the trade offs in link speed and link distance. Additional items were jitter associated with the 10-wide interface. Questions from the floor on how much extra distance can be gained by controlling the filling of the fiber modal bandwidth. Stan Swirhun (Vixel Corp) gave a presentation on Optical Power Budget at 1.25Gb/s. Major issues addressed were, issues associated with OFC (open fiber connection), TIA underestimation of the Mulit Modal Bandwidth, experimental tests at OC24 (1.244G) vs FC, discussion on optical budget issues. and a proposed specification for the optical connection, conclusion is that 1250Mb/s at 500M on 62NMF is feasible with some work. Pat Gilland, (Methode Electronics) gave a presentation on Optical power budget for CD-Lasers. Major issued addressed were specifics of ethernet optical requirements, results of experiments performed at 1.06 and 1.25Gb/s on 500m and 1Km. Questions on why 8B10B coding was selected for FC (mostly influenced by data run lengths and other items). Howard Frazier ask for a show of hands on how many people plan to participate in the standards process (81) and how many companies plan to participate (54). This information will be used to update the Broad Market Potential portion of the 5 criteria document. Allistair Black (Gadzoox) made a presentation on The Fiber Channel Jitter Issue. Major Issues addressed were sources of and absorption of jitter and what changes we may wish to make to the jitter specification to better fit our requirements while still maintaining BER >10-12. Miscellaneous experimental and empirical test results were presented and discussed. Sailesh K. Rao (SDE) made a presentation of 1Gb/s Over 4- Pair - Preliminary Simulation Results. Major issues addressed were a Split Passband Proposal for operation over 4 pair category 5 wire. This presentation differed from his previous proposal in changing to a 9 level signaling from 8 level. Supporting calculations and considerations were included. Questions about what the latency and synchronization and how they relate to the work in the T2 WG. There are still a number of issues that need to be investigated further. Questions about cost are expected to be 2X that of T2. Michael Jay Lieb (Technitrol) made a presentation on Gigabit Transmission Over Copper Media. Major issues addressed were phase frequency distortion present in twisted pair, issues associated with eye patterns, along with numerous example of eye patterns under various conditions. Included in the presentation were some results of eye patterns at 1.125Gb/s on UTP Cat 5. Question on output voltage was 800mVpp. Conclusion is that category 5 cable can support a 2 level, 25 meter cable can be achieved. Question on what the fixed optimized cable length was, 150m. Minutes of Milpitas,CA (Jan 11 & 12) meeting distributed. Richard Taborek (Amdahl) made a presentation on Use of SOP/EOP codes, and OFC. Major issues addressed were a how the various signaling aspects of Fiber Channel would be applied to ethernet and where some minor changes to some of these would benefit the HSSG effort. Issues associated with 8B10B coding were addressed in detail along with OFC operation. Additional discussion on synchronization, auto negotiation, link issues and various delimiters. General discussion on how and when auto-negotiation occurs followed with issues associated with RJ45’s and fiber applications. Also the issue associated with 8B10B symbols and the auto- negotiation sequence. Andreas Bechtolsheim (Granite Systems) made a presentation on 8B10B Coding and Delimiters for 1000Mbit/sec Ethernet. Major issues addressed were what the minimum subset of the 8B10B coding table would be useful for use by ethernet. Special focus on delimiter size and implementation. Discussion on DAP characters and Fill roles and utilization. Howard Frazier adjourned the meeting after re-iterating the schedule for tomorrow. Continuation of HSSG meeting La Jolla, CA 3/13/96 Howard Frazier brought the meeting to order at 1:00pm The agenda for the afternoon was presented and discussed: 1) Continue Technical Agenda 1:00 - 3:00 2) Review Objectives % PAR & 5 Criteria 3:00 - 4:00 3) Respond to other WG Comments 4:00 - 4:55 4) HSSG position on PAR & 5 Criteria 4:55 - 5:10 5) ANSI Offer of Cooperation 5:10 - 5:30 6) Cooperation with 802.12 5:30 - 6:00 7) Sub Task Force Organization 6:00 -6:15 8) Plan for Interim Meeting 6:15 - 6:25 9) Affirm Minutes from Jan Interim Mtg 6:25 - 6:30 10) Adjourn 6:30 Howard noted that item #3 must be completed prior to 5:00pm as required by 802 so presentations must stay on schedule. Jonathan Thatcher (IBM) made a presentation on Thoughts on gigabit Ethernet Physical . Major issues addressed were a brief history of the Fiber Channel developments continuing with background information on OFC, Link Acquisition, and Coding. Additional material on common problem areas was supplied. The conclusion was to suggest that HSSG adopt the existing Fiber Channel operating speed. Western Digital maintains an FTP site that contains the spec for the 10B Coding. Howard Frazier (SUN) made a presentation on Scaling CSMA/CD to 1000Mb/s, Experimental Results. Major issues addressed were experimental testing using 100BASE-FX over long lengths to approximate the shorter lengths that 1000Mb/s would be used in. Conclusions were that increasing the slot time and carrier extensions have a minor adverse effect on throughput for packets greater than minimum size. Howie Johnson presented two foils obtained from David Allen and Richard Prentice of TI detailing contacts for the more information on Fiber Channel. Fiber Channel Association phone number is 1-800-272-4618 X3T11 Chair Roger Cummings Distributed Processing Technology 140 Candace Drive Maitland, FL 32751 Phone: 407-830-5522 X348 Fax 407-260-5366 Email: cummings_roger@dpt.com Fibre Channel ftp sites: 10 Bit Technical Report ftp://fission.dt.wde.com/pub/standards/10bit/postscript Fibre Channel Reflectors TO:majordomo@
, EX: (majordomo@think.com) Suscribe fibre-channel-ext@think.com 10bit-owner@dt.wde.com stds-802-3-hssg@mail.ieee.org ANSI Documents Global Engineering 800-854-7179 WWW http://www.amdhal.com/ext/CARP/FCA/FCA.html http://www.cern.ch/HSI/fcs/fca.htm http://www-atp.llnl.gov/atp/telecom.html Mohan V. Kalkunte (AMD) made a presentation on Scaling CSMA/CD to 1000Mb/s, Simulation Results. This presentation was a continuation of Howards presentation based on simulation results. The conclusion reached was that simulation model results match closely with experimental results achieved in Howards presentation. Mart L. Molle (UC Riverside, CS Dept) made a presentation on Reducing the Effects of Propagation Delay on CSMA/CD Networks. Major issues addressed how to reduce the “cost” of collisions in ethernet. Several techniques were presented and discussed pointing out merits and drawbacks. Conclusions drawn were that incoming collision deletion could be incorporated and maintain computability with standard CSMA/CD DTEs. This approach requires all the repeaters within a single collision domain to implement this functionality. Discussion and several questions followed the presentation. Ahmad Nouri (Compaq, IPG) made a presentation on Gigabit Ethernet Survey. Major issues addressed were the results of a market survey commissioned to investigate the interest level in gigabit ethernet. Parameters of the survey were covered along with the results of the survey. Conclusions of the survey indicated that interest in 100Mb/s was growing very rapidly and 86% of the respondents expressed interest in GB Ethernet after hearing about it. Ahmad agreed to put the original questions ask in the survey on the HSSG reflector for all to inspect. Wayne Rickard’s presentation on ANSI offer of cooperation was deferred until later. to allow time to address the responses to the PAR and 5 Criteria. Howard Frazier presented the Proposed Objectives. Discussion on the stated objectives were generally confined to clarification and minor commentary with no specific objections detailed. Bernard Daines did however ask about the wording associated with Fibre Channel suggesting that some clarification of our intent with respect to existing Fibre Channel components. Howard Frazier summarized that FC was the basis. Another item questioned was associated with the objective of 25m of copper pair. The questioner believed that 100m should be the goal. Item #1 was discussed in greater detail. Jonathan Thatcher suggested that accepting Fibre Channel as is in the best interest of the HSSG. Additional commentary continued on the pros and cons of working for the 1.25G speed vs the 1.063G speed. A straw poll on 1.0625G vs 1.25G rate was taken 1.0625G 17 1.25G 38 Abstain 6 Item #11b was addressed next. Howard took commentary from the floor on the subject of if the decision on of 25m or 100m should be addressed now or later in the task group. Most commentary concerned the inclusion of 25m being a valid solution or goal of the group. A straw poll was taken on: Do you want to decide between 25m and 100m copper links now or wait until we form a Task Force. Now 7 Later 56 Item #13 regarding IEC/ISO 11801 was discussed and Geoff Thompson gave a brief summary of the contents of the document. Motion by Shimon Muller Second by Bernard Daines Reaffirm Proposed Objectives as written Y 61 N 0 A 3 The motion passed Discussion then turned to the 5 criteria. Howard ask for comments and concerns on: 1) Broad Market Potential -No response. 2) Compatibility with IEEE Standard 802.3 - No Response. 3) No Response however it will need to be addressed later. 4) Technical Feasibility - Gadi Lahat ask for a clarification on text added Sunday night (underlined text) 5) Economic Feasibility - Howard indicated that a previous comment will need to be addressed later. The PAR was presented: No comments on page one Howard suggest the paragraph 7 may need some additional revisions to better focus the wording. Howard also displayed the letter from IBM indicating that they are willing to meet the IEEE requirements regarding licensing of patents that may be included in the HSSG effort. Paragraph 9b was read and explained also. Specific Comments from other working groups were addressed next. Responses to Comments received by the 802.3 HSSG regarding the 802.3 Gigabit PAR. Comment: Gigabit Ethernet needs priority. The lack of priority is a flaw in 802.3 that will become fatal at an operating speed of 1Gb/s, due to the need to aggregate multiple lower speed links together and support multiple services in a backbone application. Priority should be addressed in the gigabit PAR, and not left for separate work item. Response: The 802.3 HSSG does not agree that the lack of a priority mechanism constitutes a fatal flaw in the CSMA/CD access method regardless of the operating speed. The 802.3 HSSG believes it would not be appropriate for this study group to further investigate this subject. Comment: Need to define another operating speed (100< s < 1000), because there will be a significant cost advantage for the sub-gigabit Ethernet, particularly when the switch and bridge cost/bandwidth trade-offs are considered. Response: The 802.3 HSSG believes that it is not constructive to standardize multiple operating speeds with in the same power of 10Mb/s because this would impair multi- vendor interpretability. The HSSG also believes that the cost advantage of a sub-gigabit system will diminish over time. The HSSG believe that experience with ATM and Fibre Channel validate this position. Comment: Since CSMA/CD cannot be used in networks of useful size, there is no distinct identity since Gigabit Ethernet = Full Duplex = 802.12 Response: Several presentations have been given to the HSSG which demonstrate that CSMA/CD at 1000Mb/s can be used in networks of useful size, and the HSSG has established specific objectives regarding the continued support for CSMA/CD within 802.3z. The 802.3z project is definitely not limited to Full Duplex operation. Comment: Gigabit Ethernet is not needed since there is going to be a new project to develop a new MAC that will coalesce everything into one standard. Response: No presentation have been made of which HSSG is aware concerning this “new MAC”, and the 802.3 HSSG believes that the virtues of the CSMA/CD MAC have been amply demonstrated in the market place. Comment: There should be one and only one 802 standard with an operating speed of one gigabit. Response: The 802.3 HSSG firmly believes that there is a very large user community which will be best served by a gigabit supplement to the 802.3 standard, and that this sufficient justification for this project. This assertion is well supported by the data provided in the response to the “Broad Market Potential” criteria supplied with the 802.3z PAR, and by market survey data presented to the HSSG. Comment: The second paragraph of “3. Distinct Identity” implies that there are 60 million 10/100 Mb/s agile CSMA/CD nodes in existence which is untrue. Response: Strike 10/100. Comment: The last sentence of “5. Economic Feasibility” implies that the cost model for copper horizontal cabling runs has not been established, which is untrue. Response: change the sentence to read. “while the cost model for the horizontal copper cabling is well established, the cost model for 1000Mb/s physical layers which will operate on horizontal copper cabling has not yet been firmly established. Presentations have been given to the HSSG which suggest a cost multiple of 2X relative to 100BASE-T2. Comment: As drafted, the PAR scope would allow a total revision of the MAC operation. It does not even require that the MAC defined be CSMA/CD. The scope should be altered to align it to the intent described under “2. Compatibility.” Response: Modify first sentence of “Scope” as follows. [Define] carrier sense multiple access with collision detection (CSMA/CD) media access control (MAC) [parameters and] minimal augmentation of its operation. Comment: The study group has not currently decided between using a limited distance (30-50m) or modifying the MAC to increase the network diameter (200m). If the MAC is not changed, there has been no data presented to show that Market Potential of such a limited distance shared LAN. It is not even clear that such a limited distance fulfills the functional requirements. On the other hand, there has been no examination of the impact on performance of the changes required to support the 200m distance. In shared mode, the performance increase could be drastically less than 10X. It should be shown that either the 200m network can provide a reasonable performance or that there is sufficient market for the small network. Response: Multiple detailed analyses of CSMA/CD performance as presented on 3/13/96 from credible technical sources have convinced this committee that operation at a 200m diameter is feasible, and the characteristics of such a network are understood. Comment: 802.12 primary concern is in the area of distinct identity. Distinct identity should be focused on solutions to distinct problems, not distinct solutions to the same problem. the current distinct identity presentation with this PAR does not address what distinct user problem it is solving. Of particular concern is the high overlap between this and 802.12a. Both will cover full duplex operation over Fibre Channel based PMDs. This concern could be addressed by cooperation to product a single specification covering the PMD, auto-negotiation, and full duplex operation. Response: The 802.3 HSSG appreciates the concerns expressed by the 802.12 WG and will address them more fully during the closing 802.3 plenary. Comment: Some of the objectives and technical choices do not appear to be consistent. The distance objective is 500m and the statement “Customers in many cases be able to re-use their existing fiber that has been installed in accordance with ISO/IEC 11801.” is made. However, the current distance on 62.5 micron fiber for Fiber Channel is 300m. The speed increase will drop that distance to 230m. While 50 micron fiber would support a distance closer to 500m, the vast majority of currently installed fiber in the US and Europe is 62.5 micron. No data has been presented to indicate the distribution of distance in installed fiber. Increasing the speed at the expense of distance on 62.6 micron fiber is a disservice to the user. Users may benefit more from being able to use their existing cable than the 25% greater data rate. Response: The objective of 802.3z is a 500m span over multi-mode fiber. If during the life span of the 802.3z task force it is proven that this goal in not achievable, this objective will be modified. Task Force objectives lists are not part of the PAR application process. Comment: Further, we are concerned about the technical feasibility of the speed increase. The Fibre Channel committee has been struggling with jitter issues on the FC-0 interface at the current data rate. They have just recently worked out a compromise. While several transceiver suppliers have indicated an ability to create higher speed transceivers, there has been no indication that the chips on the other side of the MII can tolerate the effect of the speed increase. The Study Group has not reviewed the FC-0 interface, nor any other Fibre Channel specifications to determine the impact of the increase. Response: Even a modest decrease in the rise/fall times of the transceivers and chips used in gigabit links will substantially improve overall jitter performance. Over the expected life of this standard it is entirely reasonable to anticipate such developments, particularly in light of the act that the 802.3 project will not have to contend with certain installed base compatibility issues that may have tied the hands of the ANSI committee. Furthermore, the Fibre Channel Jitter Study group was created to deal with the issues created by the Fibre Channel Arbitrated Loop topology. This loop topology is not applicable to 802.3. Howard took a straw poll concerning those in favor of writing PAR for a project with a range or specific operating speed. Range 23 Specific 31 Abstain 2 Motion: Shimon Muller Accept the Responses to comments as presented and forwarded to 802 WG chairs. Second Colin Mick. Motion passed by acclimation The purpose (paragraph 7) is modified to “ ... installed base of 10/100 CSMA/CD nodes previous investments in research and development and principles of network operation and management. Motion: Rebecca Farley Motion to Affirm the PAR as modified to 802.3 for approval Second Esser Y 48 N 0 A 0 Motion: Colin Mick motion to affirm 5 criteria as modified and forward to 802.3 for approval and forwarding to exec. Second Daines Y 52 N 0 A 0 Wayne Rickard (Emulex) made a presentation on ANSI X3T11 Liaison Report. Issues covered were on the ANSI groups interest in working with the HSSG in the development of a standard. The study group expressed interest in establishing a liaison to ANSI X3T11 but no one volunteered to act as such a liaison. (Continuation of discussion on liaison with ANSI Fibre channel group, minutes taken by David Law) Geofff suggested that if a liaison is formed approval would have to go up to the executive. Geoff then went on to describe his experience of being a liaison. A discussion then ensued on the need for a liaison. A quick show of hands on the need for a liaison showed approx. 25 for, 0 against. Howard then asked for any volunteers, there were none. Howard said he would ask again in the 802.3 plenary on Thursday. Co-operation with 802.12 Howard stated that an interest had been expressed by 802.12 in the HSSG work, and in particular Full Duplex, as there may be significant common ground. He and Geoff were approached by some members of the executive committee including the chair of 802.12 and they worked together to draft a "Statement of Principle". Howard then put up the slide containing the Statement of Principle (see attachment ####). ##### Need to add slide ##### There was then much discussion about the statement, issues raised included what to do with this statement and what exactly the statement meant. Did it mean joint authorship, maybe joint voting?. Also would the final document be a 802.12 or a 802.3 document and was it co-operation on a common full-duplex PHY or was it more fundamental. Shimon Muller then proposed to table the Statement. A discussion the ensued around the fact that if the HSSG PAR was approved at the executive meeting on Thursday evening the members of the HSSG group would form a task force. The HSSG would then never meet again and the Statement could never be taken off the table. It was concluded that this would be acceptable as the Statement would be considered again in the 802.3 plenary on Thursday. Ian Crayford asked who is eligible to vote. Howard stated that this is a HSSG vote, it is procedural, the 50% rule applies. The vote was then taken. Moved: Shimon Muller Seconder: Gadi Lahat Yes:50 No:16 Abstain:4 The Statement was tabled. Sub Task Force Operation Howard Frazier presented a slide with a proposal for the sub task force arrangements (See attachment ####). Howard stated that in the past sub task forces have worked well because of the parallel work that is performed. There was then some discussion of the detail of the slide. ##### Need to add slide ##### Plan for Interim meeting Howard Frazier discussed the arrangements for the next interim meeting. He stated that an arrangement similar to the January Interim meeting, when Full Duplex and T2 also met in the same week and at the same location, would be preferable and that Rich Seifert has suggested the week of the 20th of May. So far there was one offer to host the meeting. Details would be finalised and then presented at the 802.3 plenary on Thursday. Confirmation of last meeting minutes Confirmation of the last minutes was moved and seconded, they were approved by affirmation. Adjournment Adjournment of the meeting was moved and seconded, this was approved by affirmation.