January 23, 1990 TO: X3T9.3 Fiber Channel Working Group Members FROM: Roger Cummings SUBJECT: FIBER CHANNEL WORKING GROUP MINUTES Please find attached a draft of the minutes of the ANSI X3T9.3 Fiber Channel Working Group of January 11 and 12, 1990. Note that there are also twelve Attachments to the minutes that relate to presentations at the meeting. A NASA Technical Brief named "Using Bit Errors to Diagnose Fiber Optic Links", and an extract from the Protocol Engines "Transfer" magazine on the subject of Multicast are also included as an Attachments 13 and 14 respectively. The next Fiber Channel Working Group will be held on February 21 as part of the February X3T9 Plenary week that is being hosted by Motorola at the Wyndham Southpark hotel in Austin, TX. A meeting notice for this week is attached along with a preliminary notice for the March Working Group week and a schedule of X3T9.3 meetings (both plenaries and working groups) for all of 1990. If there are any corrections required to, or omissions noted from, the minutes I can be reached as follows: Phone: (303) 673-6357 (Business) (303) 665-0761 (Home) Internet: 70106.307@compuserve.com Fax: (303) 673-5891 Regards Roger Cummings Senior Engineer Subsystems Controller Development MD4271 #neu0/rc 1990 SCHEDULE OF MEETINGS DATES EVENT LOCATION 01/08 - 01/12 Working Groups San Jose, CA 02/19 - 02/21 Plenaries and Working Groups Austin, TX 03/05 - 03/09 Working Groups Irvine, CA 04/23 - 04/25 Plenaries and Working Groups St. Pete., FL 05/07 - 05/11 Working Groups Providence, RI 06/18 - 06/20 Plenaries and Working Groups Wichita, KS 07/09 - 07/13 Working Groups Rochester, MN 08/20 - 08/22 Plenaries and Working Groups Seattle, WA 09/04 - 09/07 Working Groups Denver, CO 10/15 - 10/17 Plenaries and Working Groups Valley Forge, PA 10/29 - 11/02 Working Groups TBD 12/03 - 12/05 Plenaries and Working Groups San Diego, CA The format of the week for months containing only working groups is as follows: Monday SCSI Common Access Method Group Tuesday and Wednesday SCSI Working Group Thursday and Friday Fiber Channel Working Group The format of the week for the months containing both plenaries and working groups is as follows: Monday Separate X3T9.2 (SCSI) and X3T9.3 plenaries (Monday portion of X3T9.3 devoted to IPI and HPPI) Tuesday Separate X3T9.2 (SCSI) and X3T9.3 plenaries (Tuesday portion of X3T9.3 devoted to Fiber Channel) Wednesday Fiber Channel Working Group. DISCOUNT AIRFARES The following discount airfares are available for travel to and from both the February plenary week and the March Working Group Week. These group schemes typically offer a 40% discount on the Y class fare (with no additional restrictions), or a 5% discount on any another published fare. Arrangements must be made at least seven days in advance with the Meetings desk (see special 800 number below), but ticketing is performed by you agent as normal. AIRLINE ACCOUNT NUMBER RESERVATION PHONE GROUP NAME United 433FF 1-800-521-4041 Accredited Standards Committee (ASC) Northwest 15445 1-800-328-1111 ENDL Please use these special fares as they will be discontinued by the airlines if not used MARCH WORKING GROUP WEEK The March Working Group Week is being hosted by Western Digital from March 5-9, 1990 at the: Red Lion Inn 3050 Bristol St. Costa Mesa, CA Please call the hotel directly at (714) 540-7000 for reservations. A special group rate of $75 single and $80 double has been arranged. The group name is WD/SCSI and the cutoff date is February 28, 1990. As usual, the following meetings will be held: Monday SCSI Common Access Method Group Tuesday and Wednesday SCSI Working Group Thursday and Friday Fiber Channel Working Group MINUTES OF THE FOURTEENTH FIBER CHANNEL WORKING GROUP MEETING The Fourteenth meeting of the ANSI X3T9.3 Fiber Channel Working Group was at the Red Lion Hotel in San Jose, CA on January 11 and 12, 1990 as part of the January Working Group week hosted by Bob Snively of Sun Microsystems. A total of 58 people attended as follows: ADVANCED MICRO DEVICES Jim Kubinec Yun-Che Wang AMDAHL Ron Kreuzenstein Rich Taborek AMDAHL/KEY COMPUTER LABS. Ed Cardinal Bob Cargnoni AMP Charles Brill Bob Southard ANCOR COMMUNICATIONS Dick Everson Bill George Jay Swenson BT&D TECHNOLOGIES Ray Johnson Jim Somerville CANSTAR Kumar Malavalli Douglas Mitchell CONTROL DATA Wayne Sanderson CRAY RESEARCH INC. Marvin Bausman Wayne Roiger ENDL Dal Allan ESL INC. David F. Hepner FORD AEROSPACE Gary Waldeck FUJITSU AMERICA Bob Driscal Kohji Mohri GAZELLE MICROCIRCUITS David MacMillan HEWLETT PACKARD Del Hanson IBM Richard Bates K. C. Chennappan Charlie Martin Joseph R. Mathis Gerald L. Rouse Frank Shott Ron Soderstrom John Sorg Daniel Stigliani Horst L Truestedt Carl Zeitler IPITEK INC. Michael Pugh LASERTRON Jim Lewis LAWRENCE LIVERMORE NATIONAL LABS. Paul Rupert John Severyn LOS ALAMOS NATIONAL LAB Michael McGowen Richard Thomsen Don Tolmie January 11&12/90 Fiber Channel Working Group Minutes Page 2 MAXTOR CORP. Dave DeLauter Paul Guantonio Larry Lamers NATIONAL SEMICONDUCTOR Sam Laymoun NCR John Lohmeyer ORTEL Alexander Leibovich PCO Jim Goell SHELL DEVELOPMENT CO. Patric Savage SIEMENS FIBER OPTIC COMPONENTS Schelto Van Doorn STORAGETEK Roger Cummings Floyd Paurus SUMITOMO ELECTRIC Harry Takada SUN MICROSYSTEMS Noel Lindsay Robert N Snively TANDEM COMPUTERS Mary Henske The meeting was opened by the Chairman of the Working Group, Dal Allan of ENDL Consulting, who presented a brief agenda and a summary of the Fiber Channel characteristics. A copy of Dal's slide is Attachment 1. The first presentation was made by Jim Kubinec of AMD. Jim had made a presentation at the December Working Group meeting attempting to quantify the effect of the dc imbalance of the 4B/5B code. That presentation made some assumptions about the maximum frame size and the inter-frame gap which did not correspond to the latest FC-2 protocol and which did not use worst case data patterns. Jim was therefore requested to rework his analysis on the basis of the FC-2 parameters and worst-case data, hence the presentation at this meeting. A copy of Jim's slides is Attachment 2. Jim emphasized that in "starting from scratch" it would be natural to use a dc balanced code, but that his purpose in performing this analysis was to evaluate if it was technically possible to use the existing 4B/5B parts and thus leverage off their existing installed base. His analysis indicated a Bit Error Rate (BER) penalty of at least one order of magnitude due to the characteristics of 4B/5B versus a balanced code. Del Hanson of Hewlett-Packard immediately challenged that figure on the basis that Data Dependent Jitter and the effect on receiver sensitivity of having to open the receiver eye pattern. Jim agreed that this had not been addressed in his analysis. Patric Savage of Shell Development noted that the Bit Error Rate specified is reached early in a frame, and Jim agreed but noted that an additional 1 dB of launch power regains the loss in BER. Del Hanson noted that the Data Dependent Jitter effect requires at least another 1 dB. Dal Allan questioned if these effects were significant given the 262.5 Megabaud application that was being discussed, and was told that they were. Mike Pugh of Ipitek noted January 11&12/90 Fiber Channel Working Group Minutes Page 3 that the availability of 250+ Megabaud parts may be decreasing due to increasing competition and specmanship at FDDI speeds. Schelto Van Doorn of Siemens countered by stating that such parts could be produced by special trimming during manufacturing. Del Hanson of Hewlett-Packard then described an method of link analysis that had its roots in an FFT analysis that he had performed during the design phase of FDDI. His latest version of the analysis used gaussian equations and a spreadsheet, and it was this version that he presented. Del initially presented an analysis of a 200 Megabaud link, but was able to return to his office and rework the analysis on the basis of a 266 Megabaud link, and present that analysis later in the meeting. It is that later analysis that is included as Attachment 3. Jim Goell of PCO asked Del how he had derived this analysis from the FDDI parameters, and was told that he had increased the bandwidth of the receiver, and scaled the jitter and receiver parameters directly according to the frequency. Del noted that this was not the only tradeoff that could be made and gave an example of trading output power for speed while keeping the receiver sensitivity constant. Jim Goell noted that the link length could also be limited, and Del agreed that that would help the Data Dependent Jitter analysis. Dal Allan asked Del if his analysis showed that 4B/5B coding would not operate on this link, and Del replied that he had only noted that 4B/5B would incur a 1 dB penalty. Sam Laymoun of National Semiconductor noted that a 1 dB penalty would result from peak to peak jitter alone. Schelto Van Doorn noted the importance of Data Dependent Jitter on data links, and stated that a balanced code results in an increase in receiver sensitivity. The next presentation was given by Ray Johnson of British Telecom and Dupont Technologies (BT&D) on the subject of 250 Megabaud components. Ray introduced BT&D as a manufacturer of datalinks with a considerable interest in double speed FDDI parts. He displayed three eye-patterns resulting from testing using optical attenuators and 2 kilometers of fiber, and copies of these slides are Attachment 4. He noted that the anomaly in fall times shown in the second eye-pattern was due to anomalous behavior in the transmitter. Richard Bates of IBM asked how the part that was tested related to the standard FDDI part, and was told that the modification for 250 Megabaud operation involved removing the bandwidth-limiting filter at the receiver input and a making a minor modification to the led transmitter. Marvin Bausman of Cray Research asked if there were any parametric differences in the fiber and attenuator tests, and Ray replied that the differences were insignificant because their transmitter was an eled which has a much narrower spectral width than an sled. Schelto Van Doorn asked if BT&D would be able to produce 262.5 MegaBaud parts and Ray stated confidently that they would, although a new chip January 11&12/90 Fiber Channel Working Group Minutes Page 4 would be required. Horst Truestedt of IBM then introduced an IBM proposal that would be detailed in following presentations. A copy of Horst's presentation is Attachment 5. The proposal was based on the idea of a dual-standard in which an FC-0 using longwave lasers and single mode fiber is specified for use at all distances and data rates, and an FC-0 using shortwave lasers and multimode fiber is used at a 265 Megabaud rate and for medium distances. In addition IBM would make a technology presentation on gigabit multimode medium distance links. The Longwave portion of the IBM proposal was made by Dan Stigliani from slides prepared by Jerry Radcliffe. A copy of these slides is Attachment 6. Dan began by describing the philosophy behind the IBM proposal. He noted that the International Electrochemical Commission (IEC) 825 safety standard that has the force of law in some European countries has stringent requirements, and that it was a design goal to minimize the impact of this standard. Also, he noted that for the distances involved in this portion of the proposal there is "a lot of light", and that this fact was used to design for low cost and manufacturability. A large spot size was used, and tolerances required could approach that of multimode systems - with sorting used to obtain components for the 10 Km version. Also a standard SONET class of single mode fiber was assumed. David MacMillan of Gazelle Microcircuits asked for clarification of the IEC requirements, and was told that for Class 1 lasers (the intended class) the only requirements were a small label on the box, and no special training or interlocks would be required as no laser safety hazards existed. It was pointed out that precautions had to be taken to prevent a person from looking directly into the laser, and Mike Pugh noted that it is complex to calculate the amount of light that can enter the eye in that case. Ron Soderstrom of IBM noted that while that calculation is necessary in order to meet the US Safety standards, which are published by the Center for Devices and Radiological Health (CDRH), it is irrelevant with regards to IEC 825 because of the fixed (and large) aperture size that it defines. Assuming that there is unlimited access to the laser, the only method of meeting IEC 825 is therefore to provide a means of limiting the laser output power when such an access is detected. Dan then presented Loss Budgets and Diagrams for both the 2 Km and 10 Km links. John Severyn of Lawrence Livermore National Labs. questioned the derivation of the dc loss figures in the diagrams, and was told that they represented a 3 sigma analysis. Dan gave the derivation of the number of splices as being from an algorithm of one per kilometer plus one, plus two for repairs. He also pointed out that the use of both 9 um and 10 um cable had been assumed, and that a coupling loss had be included. John Severyn noted that the latest AT&T cable is 8.3 um, but was told January 11&12/90 Fiber Channel Working Group Minutes Page 5 that the mode diameter is still in the 9 um range. Dan then gave the loss figures in the CCITT table format, and stated that the -25 dBm receiver sensitivity is achievable. John Severyn asked how the laser parameters were defined, and was told that they represented the total variation over the life of the laser and its coupling. John questioned if the -3 dBm level shown would meet the safety standard, and was told that the existence of an interlock to prevent direct viewing of the laser source had been assumed. Jim Goell suggested that the spectral width be decreased and the center wavelength range expanded, and it was agreed that the parameters could be adjusted. However it was noted that some CCITT numbers stretch to 1355 nm and that the fiber has a water peak at 1380 nm. Dan stated that the reason the figures had been chosen was to avoid the requirement for an optical isolator at the transmitter. Dal Allan noted that the only difference between the 2 Km and 10 Km versions was the receiver sensitivity, and Dan agreed that this may not be significant in terms of cost. Jim Goell cautioned that dynamic range can cost more than sensitivity and Dan agreed that 22 dBm may be difficult to meet. It was noted that this figure would not be difficult for a discrete transimpendance amplifier, but Jim Goell again cautioned that it could be difficult to meet with a monolithic design. Del Hanson noted that the FDDI methodology is to allow a laser or a led transmitter to work through the cable in to the same receiver, and Dan stated that this was not their mindset. Kumar Malavalli of Canstar emphasized that the 10 Km version was a passive link and that a possible alternative tradeoff was the use of an active repeater. It was noted that while single mode systems have a number of technical advantages they also have some drawbacks in terms of handling, alignment and environment (dust, dirt etc.). While noting that this was true, Dan stated that he had been very impressed in the advances in single mode connectors over the last two years. John Severyn noted that in some situations single mode fiber can be cheaper than multimode, and especially in terms of multi-fiber cables. Dan agreed that cost optimization has to be performed on the basis of the complete link, and not just by individual components. Noel Lindsay of Sun Microsystems questioned the number of wiring boxes implied in the budgets, and suggested that active repeaters and routers are a more likely implementation. Dan agreed that in a campus environment active repeaters could be used to save cost. However he noted that IBM has a current product (the 3044) which operates over 2 Km and is mostly used in a campus environment, and that their experience is that most installations put in many more patch panels and splices than were initially estimated. John Severyn emphasized this by stating that at Livermore Labs they January 11&12/90 Fiber Channel Working Group Minutes Page 6 have a square-mile campus, and they use active concentrators - but they still use passive patches and splices because "people move every year or so". January 11&12/90 Fiber Channel Working Group Minutes Page 7 The afternoon session began with the shortwave and multimode portion of the IBM proposal, which was presented by Ron Soderstrom. A copy that portion is Attachment 7. A number of comments were made on the selection of 50/125 um fiber for the proposal. Ron noted that 62.5 um fiber could be used but that its bandwidth.distance product is limited to 120- 140 MHz.Km while 50 um fiber can meet 500 MHz.Km @ 780 nm and 80 MHz.Km @ 830 nm. Del Hanson stated that the worst case specification for FDDI fiber is 150 MHz.Km @ 850 nm, and Ron agreed and extrapolated this to 120 MHz.Km @ 780 nm. Ron noted that the use of a self-pulsating laser with a short coherence length had eliminated the modal noise problem usually associated with multimode systems, and he stated that the rise and fall times for the laser are an order of magnitude faster than for leds. He noted that the IBM proposal includes the concept of a physical level repeater between a shortwave and longwave segments. Additionally he noted the link monitoring capability that is inherent in the laser structure. Ron stated that output power of the shortwave laser was limited in order to comply with Class 1 of IEC 825. John Severyn asked to what standard the commonly available laser pointers complied, and was told that they meet the Class 2 requirements but that these requirements were only applicable to visible light sources (i.e. less than 700 nm). Del Hanson asked Ron if he believed that the Class 1 limits would be changed due to ongoing research on eye damage from invisible lasers. Ron agreed that there is an ongoing effort, and mentioned a Japanese proposal to the IEC to limit the aperture width which would effectively bring the IEC regulations into line with CDRH. However he suggested that this is immaterial in that it is possible to meet the regulation s as they exist today. However the method of meeting these regulations uses an Open Fiber Control technique. Ron then presented a Loss Budget and an Loss Diagram for the Shortwave link. He noted that they assume that the set point for power is in the fiber instead of the receptacle, and that the laser variation is not as wide as it was the Longwave link because of more reliable connection coupling. He also noted that the center wavelength variation of the laser is constrained by the compact disc application because of the wavelength sensitivity of the recorded media. Ron then moved on to describe an IBM implementation of an shortwave laser Optical Link Card (OLC). A copy of this description is Attachment 8. The OLC is designed to operate at 220 Megabaud, and the optics are known to work at 265.6 Megabaud and have even been tested at 300 Megabaud. However it is known that the logic on the Card will not operate at 250 Megabaud over the full environmental and power January 11&12/90 Fiber Channel Working Group Minutes Page 8 supply ranges. The card incorporates the Open Fiber Control (OFC) technique - which requires a full duplex link and upon which IBM has a patent. The OLC operates from a single +5 volt supply. It uses the SC connector, which is a convex polish physical contact connector type, and IBM has seen an average insertion loss of 0.2 dB from four manufacturers. A plastic bridge mechanism is used to combine two connectors into a duplex unit, and the bridge is keyed to prevent incorrect insertion. The OLC incorporates all of the features required for safety conformance, and thus IBM were able to certify just the OLC and not the various enclosures in which it will be mounted. They have received letters from CDRH, VDE and SSE (Sweden) which confirm conformance. The only requirement on the mounting enclosure is a small Class 1 label (in addition to the one on the OLC). Del Hanson asked what changes could be made without requiring recertification, and was told that such things as a new data rate could be done on the basis of a new member of the same equipment family. Ron described the card as having a parallel bus interface to the rest of the equipment, and noted that encoding is performed off the card on the parallel data stream. The transmitter operates from the bus clock, and the receiver recovers its clock from the data stream. Patric Savage asked if the serializer on the OLC checks for legal codes, and upon being told that it did not suggested that a coding fault could therefore damage the transmitter. However it was pointed out that it is the average power that has to be controlled to ensure proper laser operation, and that the back facet detector controls the average power to prevent damage. Del hanson asked where the cut in point of the OFC was defined, and was told that it is set at -22 dBm, or 3 dBm below the dynamic range at a BER of 10**-12. Marvin Bausman asked Ron what would be required to market the OFC as a Class 3B product rather than Class 1, and was told that it would be necessary to affix a large red starburst label to the equipment, to provide an interlock to prevent access (to patch panels as well as the link endpoints), and to provide special equipment (e.g. goggles) and training for service personnel. Ron noted in closing that the OLC used entirely surface mount components except for the transmitters and receivers which were located in slots, but whose pins were soldered by surface mount techniques. He noted that the transimpedance amplifiers are covered by a metal shield, and in answer to a question noted that the shortwave lasers are used in the standard cd package, and that they cost $5 in 50,000 quantities. Richard Bates of IBM then gave a presentation on gigabit multimode links. A copy of Richard's slides is Attachment 9. He January 11&12/90 Fiber Channel Working Group Minutes Page 9 began by noting that the presentation had first been given at a conference in Houston in 1989, and that pictures of the featured Gallium Arsenide chips were to be found in the January 1990 issue of IEEE Spectrum. Richard described some link research activity within IBM, and listed the technology requirements that had been derived from such research. He noted that in telecom applications there is not the same pressure for integration as in the datacom world. He stated that the 1 um Mesfet process used is fairly standard, and that the state of the art is currently at 0.8 um. He noted that the quantum well lasers have sub-milliamp thresholds, and can be driven with no feedback. He stated that it is possible to group two lasers together so that they both shine down the same multimode fiber, and thus laser reliability is not a serious issue. He emphasized that no optical or electrical crosstalk had been detected in their prototype evaluations, and attributed this in part to the fact that the drivers operate at asynchronous frequencies. He described the four driver transmitter chip as dissipating only 400 mW. On the receiver side, Richard noted that an MSM detector was used for compatibility with the Mesfet process. He emphasized that an Indium Phosphide detector would be superior, but that it was not used because Indium Phosphide does not support the same level of integration as Gallium Arsenide, and that there are distinct advantages from the view of noise immunity in having very short connections from the detector to the amplifier. John Severyn asked if the fiber was fixed in place, and Richard replied that it was and the receiver chip was essentially a pigtail. Sam Laymoun asked how the fiber was aligned, and Richard replied that active alignment could be used as it was only a research project, and agreed that the result was fragile. As regards the fiber used, Richard noted that they had seen the cost of "top bin" multimode fiber as 36 cents/meter versus 20 cents/meter for standard single mode fiber. He noted that the modal bandwidth of multimode fiber is adequate, and stated that Philips are achieving in excess of 3 GHz.Km from their process. He emphasized that multimode fiber allows very robust systems, and looked forward to the day when fiber connectors would be treated like today's connectors on PCs. Richard noted that their third pass of the Gallium Arsenide worked and attributed this to "fiddling by the process people". He reported that a link consisting of 100 meters of fiber and an optical attenuator had run error free for 20 days. Richard closed by providing a Loss Budget and Diagram for a 1 Km, 1062.5 Megabaud link, and concluded that such links will be feasible in the near future at a very attractive cost level if a significant market can be established. January 11&12/90 Fiber Channel Working Group Minutes Page 10 Horst Truestedt then summarized the IBM proposals, and provided some slides produced by Albert Widmer on the subject of 8B/10B implementations. These slides are included as Attachment 10. Thursday's proceedings ended with a presentation by Don Tolmie of Los Alamos National Labs on possible data stream structures for the Fiber Channel. A copy of Don's presentation is Attachment 11. A considerable discussion was initiated by Don's presentation, with much of the disagreement seemingly centered around differences in vocabulary. Dal Allan objected to the use of the terms full duplex and dual simplex and substituted balanced and unbalanced for the cases where the dataflow between the fabric and the node is equally in both directions, or just in one direction respectively. Different types of fabric were defined, including a "railroad yard" where datapaths in both directions are switched together between two nodes, and a system where one node could be sending data to a second node and receiving data from a third simultaneously. This latter case requires of course that the fabric be able to multiplex the acknowledge frames from the opposite dataflow with the dataflow in each direction, and it was felt that this might require an expensive type of fabric. Don stated that the objective of his presentation was to obtain guidance to take to the Switching Subject Specific Working Group (SSWG) that is to be held in Santa F , NM on February 1 and 2, 1990. In this regard he did not achieve his objective, as the feeling seemed to be that the Switching SSWG should recommend to the Working Group the fabric types suitable for this application. The remainder of Don's presentation was then held over until the next day. January 11&12/90 Fiber Channel Working Group Minutes Page 11 Friday morning began with a presentation by Rich Taborek of Amdahl on the subject of an FC-2 Data Transfer Service Interface. Note that the slides contained in Attachment 12 are a revised version of those that Rich presented at the meeting. Paul Rupert of Lawrence Livermore National Labs. noted the concept of a single link management command set, and feared that this would prevent the work on FC-2 being completed in the near future. Roger Cummings of Storagetek asked that simple systems not be held up by a lack of definition for complex ones. Patric Savage stated that he still wanted the subject of link management to be explored. Rich also noted that the expedited frames feature in FC-2 was not used in his service interface, and suggested that they would not be used by any FC-3 variant. Dal Allan then decided that expedited frames would be deleted unless they could be justified by the March Working Group meeting. John Severyn noted the use of short frames in Rich's proposal and suggested that the identification of the actual data within a frame be an FC-3 function, and that the use of actual short frames be optional. John indicated that using fixed size frames is a KISS parameter, and stated that they had found in the creation of the network at Livermore that supporting short bursts is very costly. He suggested that for FC applications it may be more effective to waste bandwidth rather than support short frames. John's comments started a discussion on where the frame size was fixed, and a straw vote on a company basis produced a 10-10 tie between those who thought it was presently defined on a connection and on an exchange basis. Joe Mathis of IBM agreed to review the FC-2 document to clarify this subject. Two additional straw votes were taken with regards to the 25 Megabytes/s FC-0. The first, on a one vote per company basis, was on the coding scheme, and the results were 4B/5B - Nil, 8B/10B - 23, Abstain - 1. The second vote, on a two votes per company basis, was to determine support for the various Fc-0 alternatives. The results were: 62.5 um multimode fiber, led transmitter (FDDI) 11 9-10 um single mode fiber, l/w laser transmitter 19 50 um multimode fiber, s/w laser transmitter 16 Given the interest in all three variants, Dal Allan suggested that FC-0 documents be pursued for the present for all of them, and called for editors. The following individuals volunteered: 62.5 um m/m fiber, led xmit (FDDI) - Ray Johnson (BT&D) 9-10 um s/m fiber, l/w laser xmit - Frank Shott (IBM) January 11&12/90 Fiber Channel Working Group Minutes Page 12 50 um m/m fiber, s/w laser xmit - Ron Soderstrom (IBM) In addition Jim Goell of PCO was confirmed as editor of the FC-0 document for the 1062.5 Megabaud FC-0 variant, and Schelto Van Doorn was confirmed as the coordinator of all the FC-0 documents. IBM also volunteered to edit the FC-1 document. Don Tolmie then concluded his presentation that was begun on the previous day. It was suggested that asymmetric dataflows be handled as an FC-3 construct, but no clear conclusion was reached. There was clear agreement on one point, however, in that nobody wanted to add any complexity to the current definition of FC-2 for the purpose of supporting additional fabric types.