Rethinking System Issues

Duncan Penman penman at netcom.com
Mon Aug 21 13:15:58 PDT 1995


Time to regroup.

In the 6 months since a "System Issues Group" was formed under X3T10, it 
has accomplished essentially nothing.  We should either abandon the effort 
or redirect it so it is effective.

Here are my thoughts:

A. There are, from time to time, compatibility or integration issues
that relate to ATA or SCSI standards but are outside the scope of the 
standards themselves.  It is definitely a good thing for the standards
body members to meet the rest of the industry halfway in resolving such
problems.  In other words, there are some system issues worth working on.

B. A separate sub-organization (the System Issues Group) is not the way 
to deal with this.  The name gives the impression that there is some
separate set of people ready to deal with compatibility and related issues, 
and that simply isn't true.  If the function is to be viable at all, it 
needs to be handled as a core part of the work of the ATA and SCSI 
Working Groups.

C. I've talked to several people from inside and outside X3T10 about what
it takes to involve key players who are not X3T10 members but who are 
necessary to deal with host level issues (hardware, firmware, or software).
The major theme is FOCUS!!!  A single-topic meeting addressing a real
problem, with adequate leadtime and preparation by the organizers, will
draw the necessary participants.  It is also likely to be productive.

D. A couple of other consistent notes in this discussion:
   1. Limit the participants to those who understand and are interested
   in the subject.  I'm not personally quite so much a hawk, but it is
   reasonable to expect people to get acquainted with the subject before 
   a meeting, not during it.
   2. It takes a lot of time to assemble an adequate group for a technical 
   system level meeting.  It's not the case of just a single invitation; 
   you must jawbone, cajole, send reminders, juggle schedules, and build 
   the agenda based on the concerns of the participants.

E. I've heard it said several times that system issues overlap both ATA
and SCSI, so there should be a common focal point for both working groups
to bring their issues to.  That sounds good, but I haven't found any 
substance in it.  Coming at the question from the host side, SCSI host 
adapters have as much in common with ATM or VGA adapters as they do with
IDE.  IDE, with or without ATAPI, is the special case with all the legacy
baggage that must be dragged along.  So ATA and SCSI are better dealt 
with separately.


The bottom line here is that I'm proposing a new approach to working
system issues in X3T10 (and its eventual ATA spinoff).  When an issue
surfaces that needs attention, authorize a study group and go deal with
that specific issue.  Time frames should be short, typically 1 - 3 months.
The expected outcome should be clear up front; a report or recommendation, 
an agreement with another organization, whatever fits the problem.  Meetings
to be held if and when needed, not just held to fill a timeslot in a 
session of meetings.

PS: There are a couple of topics that fit this model.  One is the technical
report on Boot Considerations for Greater Than 8GB Devices whose project
proposal is in the latest X3T10 mailing.  I expect to have a draft of that
late next week.  The other is performance of Proxy Interrupt.  That may
be a non-issue; I posted some mail a couple of days back expressing that 
opinion.  But if it is a serious question, the right thing to do is jump 
on it and get the parameters and tradeoffs spelled out for everyone to see.
I'd like to get that question on the agenda as new business for the ATA 
Working Group later this week.

Regards,
Duncan Penman
IIX Consulting
penman at netcom.com






More information about the T10 mailing list