X3T9.2/88-71 June 28, 1988 Mr. John Lohmeyer Chairman X3T9.2 NCR MS21 3718 N. Rock Road Wichita KS 67226 Dear John, I regret I am unable to attend the working group in Boise because of my annual trip to Europe but I cannot let the opportunity pass to document some of my thoughts on electronic reconfiguration. o We have relatively few system integrators represented on the committee and all too often we overlook or underestimate the problems we create. We have to be especially careful when we deal with the physical configuration. o I am opposed to creating problems for system integrators, and adding a 60-pin configuration obsoletes all of the existing cable plant. This means the installed base will have to be reconfigured with special connector and cabling adapters. o The IBM proposal does not provide for differential, yet it is intended for external device attachments. Any work done on address reconfiguration has to support differential, for other system manufacturers if not for IBM PS/2. My personal analysis of the PS/2 and its Micro Channel capability indicates that over the next few years it could easily handle a large peripheral complement that would require differential. In my view, IBM is underestimating the ability of its own equipment by ignoring differential. o To handle reconfiguration properly, we are almost certainly going to need more pins and the only reasonable way to get them is to re-use the ones we already have. o Most of the SCSI silicon will be unable to handle any changes to the protocol which might provide a reconfiguration capability using the existing bus. This would be temporary. If we were to design a protocol sequence that preserves the existing 50-pin configuration through assignment of existing Grounds and/or re-use of the existing bus I suspect it would take chip manufacturers only a few weeks to respond with re-designed sample parts. o Gerry indicated that IBM is fixed upon the 60 pins. O.K. - but only 50 conductors need to be connected to SCSI peripherals. Apple has had no problem forcing a non-standard connector upon the world. We are only talking about a patch cord with a different connector on each end. o In the short time since the February decision, commitments have already been made by several companies which cannot be turned around, so any thought of replacing the existing 50-pin microconnector is anathema. o It may not be fair to pre-judge the Burndy proposal but I see no need to add a different mating interface to the standard. The tab and receptacle parts can be designed with a suitable back-end for PS/2 applications, but the committee has to make the need clear to the connector companies and IBM will have to provide the necessary dimensional data. IBM's proposal has been designed to suit a specific application. It suffers from not being generic enough to cover other environments. A year ago when the Vendor Unique signals were first suggested, I asked if they might be used for reconfiguring addresses on the SCSI bus. The response was NO. If the committee been asked to incorporate the ability to reconfigure addresses a year ago we would have had a solution by now, without any of the pressure that has been created in the closing weeks of SCSI-2 completion. It would not have needed anyone to reveal confidential information or suggest anything other than a deficiency in the interface which required correction. System integrators are the ones who will be hurt most by any change in the physical configuration. Please focus on protecting the installed base of SCSI peripherals. It will be easier to migrate to a new environment if the 50-pin connector/cabling scheme is retained. To summarize my position: Adding a 60-pin configuration is unacceptable if we can solve the problem by re-use of the existing signals. This would cause a temporary glitch in SCSI progress, not a permanent one like a new cable/connector configuration. Hasty decisions are all too often regretted. We need to consider the system ramifications of what we are getting into, and not push to make a date. Sorry I cannot be with you. I really would like to make these arguments in person. I could even get emotional. Have fun. Yours sincerely, I. Dal Allan