Conference 7.286::digital

    Peter it might help if you described what you are looking for.  I input
    your part number into VTX PRICE and got zero hits.  Is that a valid
    part number?  If so VTX does not know it...
    
    					Stan

    This part number description is "CI Start Coupler expander, 32 Nodes,
    2A + B Port Couplers, "
    
    	There is no inventory in the corporation. It appears that it is
    either new or very old. My guess is it is either a specially built
    product or the part number never was built.
    
    

This item has been discontinued for a few years now as can be seen in the 
SPEZKO::CLUSTER notes conference.  I suspect that the author knows this and 
is just looking to see if they can find anyone who may have one of these 
gathering dust or knows of a customer who has one that is gathering dust.  
I almost all cases there are other ways to configure the cluster to 
eliminate the need for this item.

Jilly

Peter:

  Convince them to switch from a an expanded CI cluster to a FDDI cluster?

                                   Atlant

    Further investigation yields that it was built in Albuquerque back
    in 1987. Cost information has not been updated since June 1990.
    It also appears that it might have been made in Burlington Vt. as
    well.
    
    Here is a pointer.
    
    	Call Sandra Renda at DTN 226-6174
    
    

    I sent a note to Peter suggesting he order a B2-CISCE-AA.  This is a
    refurbished CISCE-AA from Digtital Refurbished Group.
    
    Regards,
    Steve

    Hi, thanks all for your replies.  This is a CI extender that was
    discontinued a few years ago.  We haven't been able to find any and the
    reply speculating that I am looking for one collecting dust is correct. 
    Trouble is, they are too expensive to be collecting dust and the owner
    is certainly using it.  It's a shot in the dark.  
    
    Reconfiguration is always an alternative.  Most reconfiguration methods
    are much more expensive -- cost of an additional CI adapter, etc.
    
    If you do find one, even a broken one I would take it off your hands.
    
    Peter

                 -< This conference is COMPANY CONFIDENTIAL. See #1.3 >-
     Topic  Author               Date         Repl  Title
    > 1248  WILARD::HULLEY       21-DEC-1989     1  adding a CISCE needs CI utiliza
      1446  SWAPIT::ZIELINSKI    23-MAR-1990     4  24 NODE NO CISCE
      1535   STU03::APPL         27-APR-1990     6  CISCE - CI780.BIN rev20 or not?
      1627  JOCKEY::WOODWARDA    12-JUN-1990     6  CISCE Information
      1630  VCSESU::KC135Q::JOHN 13-JUN-1990     1  How many customers haveCISCE's
      1827  TRCA03::JOHNSTON      2-OCT-1990    17  CISCE and MAX_NUM_VAX ?
      2025  HACMAN::HACK         13-JAN-1991     3  CPU nodes on theCISCE??
      2135  AYOV24::LDOCHERTY    22-FEB-1991     2  CISCE ?????
      3778  SMOGGY::CAROLLA      26-AUG-1993     6  Node  placement withCISCE
      3877  SEDSWS::WOODJETTS    29-NOV-1993     2  CISCE need to be 100% sure of f
      4267   DV780::MATSUSHITA   19-SEP-1994     1  Larger cluster without CISCE-AA
      4381  TKOVOA::NARUSHIMA    12-DEC-1994     1  What is the migration model for
      4405  KYOSS1::ANDERSON      6-JAN-1995     3  CISCE cable help
      4439  NCMAIL::RIBNER        1-FEB-1995    12  Critical need for CISCE-AA
     End of requested listing
    

            <<< SPEZKO::NOTESPUBLIC:[NOTES$LIBRARY]CLUSTER.NOTE;2 >>>
                              -< + VAXclusters + >-
================================================================================
Note 4439.1                Critical need for CISCE-AA                    1 of 12
NPSS::JOHNSON "Mike J., Network Products Support"    10 lines   1-FEB-1995 13:26
--------------------------------------------------------------------------------
The CISCE is a RETIRED product.  It was removed from the proice book because
the entire remaining inventory was declared "surplus" when the WMO plant began
shutting down about two years ago, and all the CISCE's in stock there were
SCRAPPED !!!

You MIGHT be able to locate a used one somewhere internal to DEC, but the most
recent problem is that the cables also are no longer available and this is NO
suitable substitute for those.

Sorry.....

RE: .2

	It's a STAR coupler. Not Start.

							mike

    STAR COUPLER is built here in NIO. I'm not too sure of the part number
    but I know it was part of the transfer from Albuquerque to Salem, N.H.
    Being built in the SBU business.
    
    just trying to help

:    STAR COUPLER is built here in NIO. I'm not too sure of the part number
:    but I know it was part of the transfer from Albuquerque to Salem, N.H.

  The SC008 is the star coupler, and this (passive) device is available.
  Up to two of these SC008 devices can be connected, providing up to 16
  nodes in a single CI star.

  The CISCE was the (active) star coupler extender, and this device has
  not been available for quite some time.  The CISCE allowed up to 32
  nodes to be connected to a single CI star, by allowing four SC008 star
  couplers to be connected together.

  As the CISCE is no longer available, one is now required to use multiple
  CI controllers to connect more than 16 CI nodes into a CI cluster.
  (With the speed of recent systems and recent CI adapters, using multiple
  stars is usually a good idea in any event.)

RATHOLE.

This subject has been beaten to death, both here and in the CLUSTER conference.
The CISCE isn't available anymore, and indeed hasn't been for a couple of years
now.  Can we all spell OBSOLETE TECHNOLOGY ???  Further, as also discussed far
too many times, the CISCE does NOT increase the CI bandwidth, and as Steve
pointed out in the previous response, "with the speed of recent systems" it may
not be wise to put too many fast systems on the same 8.75MB bus.

Now, for those of you who want a REAL rathole.....

	The CISCE was built ONLY ONCE as a 250 unit build in Burlington,
	Vermont.  We only sold something like 16 units the first year so 
	the inventory was held in BTO until that facility was shut down, 
	then transferred to Westminster (WMO), and finally what was left, 
	beyond what MCS Logistics wanted for spares, was scrapped.
	Contrary to previous replies, it was NEVER an ABO product.

	Conversely, the SC008 was built several places, starting in Salem (NIO),
	then moving in sequence to Franklin (FXO), Burlington (BTO), Albuquerque
	(AB0), and most recently back to Salem.  We engineering types went a
	bit overboard with part numbers just to add to the confussion.

	SC008-AA, a SINGLE path 8 node star coupler panel (SC008-AB) installed
	in a low-boy cabinet.  I'm not sure why we ever offered this option 
	since we don't support single path CI.  About the only good excuse might
	be for the truely paranoid who would buy two (one for each path) so 
	that a maniac with a chain saw couldn't take out both paths at once.

	SC008-AB is the actual "panel", each one of which supports a single
	path of eight nodes.  It is a passive device (i.e. requires no power)
	but it does have electronics components, mostly in the form of
	resistors and transformers.

	SC008-AC is the "traditional" Star Coupler, consisting of two SC008-AB
	panels (one for each path) mounted inside a low-boy cabinet. 

	SC008-AD is a pair of SC008-AB panels plus the necessary hardware to
	add them into a SC008-AC for the purpose of making a 16 node, dual
	path Star.

	And finally, there WAS the CISCE.  The CISCE consisted of TWO active
	component amplifier assemblies, one for each path, plus enough cables
	to configure two sets of three or four SC008-AB's into a pair of 24 or
	32 node single path Stars.  It was ASSUMED that this would MECHANICALLY
	consist of two SC008-AC's, each with a SC008-AD.  However, the physical
	configuration could suit the engineer just so long as he did not violate
	the cable length restrictions.  I personally preferred one cab for path
	A and the second cab for path B since the cross cabling was a bear.

	I even designed a SC032 with all eight panels and both CISCE amplifier
	boxes in a single 60 inch high cab.  Fortunately, product management
	decided that it would probably not sell well after they saw the size
	of the trunk of cables coming out the bottom (about a foot in diameter).

I refuse to believe that ANYONE might possibly want to hear any more of this
foolishness so maybe the Moderator will be kind and write-lock thee note
PLEASE ?????

/mj

Hi Mike,

Bank of America, Bankers Trust, Bank of New York, Citibank.
Do these banks ring a bell? For your information they are all
in Downtown Manhattan and THEY ALL USE CISCE-AA. I just wish 
that you were a MCS Engineer and someday "preferably on New
Year's Eve" get a call from one of these banks stating that
they lost PATH "A" or PATH "B" or even better..BOTH PATHS.

For months our Logistics has been trying to find a CISCE-AA 
and keep it in stock, just for that unavoidable moment when
one of these CISCE-AA goes down, but so far we haven't been
successful in finding one. Our Managers and Engineers tremble
when they hear the word CISCE, and you said to forget about
it! If Digital made it, marketed it and sold it, then we 
SHOULD SUPPORT IT!!!!!!!!! I don't know and don't want to 
find out, what would happen when one of the above banks
finds out the we don't have a spare CISCE anywhere. This
issue must be solved before we embarrass ourselves, so
please Mike if you know where we can find a CISCE let all
of us know through these notes. Thank you, Emil

P.S. My customer, Bank of America purchased two CISCE-AA
on March 30, 1994, through Pioneer from Digital in
Merrimack NH. And now there is no more in the whole world???

Emil,

Don't get me wrong - I know you are in a bind.  But, nasty as it may be, you 
are stuck with it.  My response as shown in .12 is FACT.  I know - I was the
last Engineering Manager responsible for the product.  One of my last acts
in that function, circa, 1992 was to convince MCS Logistics to stock a 
reasonable quantity of spare BEFORE the product was removed from the price 
book.  Obviously they ignored that advice, or worse yet, someone sold the 
spares as revenue product.  For whatever reason, and despite that fact it came
out of the price book in very early 1994, people kept selling it into critical
accounts and did NOT arrange for spares. 

Now there is a good news/bad news story to go with this.  The BAD news is that 
the 70 class amplifiers used parts that are no longer manufactured by ANYONE in 
the industry.  Short of a toatl redesign, there is no way to build anymore of 
'em.  The good news is that the box has a MTBF in the high 6 digits !!!  It
draws less than 7 watts so you can guess how few active components its got !!!

The point I keep trying to make is simply this.  PLEASE, PLEASE, PLEASE stop
asking for CISCE's for NEW instalaltions.  This is SUICIDE.  If you can't 
support the one's you've got, STOP TRYING TO SELL OR GIVE AWAY MORE !!!

/mj

Write-locking this note won't make the problem go away.

I hear more and more requests from customers (and their sales/support folks)
who need a CISCE or the equivalent to allow more than 16 nodes on CI.  They
don't necessarily need 32 nodes yet, but they *desperately* need that 17th
node.

The old HSCs supported up to 48 disks each, and the new HSJ40s only 36 disks
each.  A configuration with an equivalent number of disk spindles now takes
more CI nodes than before.  With the increased I/O demands of Alpha, even more
disk spindles than before are often needed to meet the increased workload (disk
capacity is not the issue in this case -- it's requests per second).

Given that a new CI adapter for our PCI systems has been pre-announced all over
creation, the number of nodes per CI is even more likely to increase.  We
_need_ a replacement for the CISCE.  Is there anyone within Digital who is in a
position to solve this problem?  Could the same partner doing the CI adapter do
this for us?  The volumes wouldn't be high, but the margins could be quite
attractive -- we're talking about desperate people here, where the alternatives
are expensive.

Why isn't using multiple star couplers an option?  In some cases it is.  It
does require adding another CI adapter to every node (a $16K CIXCD per node
adds up quickly, plus $17K for the additional 16-node star coupler itself), or
else a portion of the CPU and I/O capacity on those nodes which _do_ have
two CI adapters must be devoted to MSCP-serving disk and/or tape traffic
between the two star couplers.

Even in cases where HSx controllers are spread across multiple star couplers
and this takes the MSCP I/O workload off the main star, it is still convenient
to have one common star coupler for the lock manager/connection manager
traffic, which is typically not nearly as heavy as the I/O workload.

FDDI allows more nodes than CI, but the performance differences with using FDDI
in place of CI make that option unattractive also.  There's also the issue
of having to replace their brand-new HSJs with HS2xx FDDI servers or something.

> Can we all spell OBSOLETE TECHNOLOGY ???

  Yes. "C-U-S-T-O-M-E-R  D-I-S-S-A-T-I-S-F-A-C-T-I-O-N"


> Don't get me wrong - I know you are in a bind.  But, nasty as it may
> be, you are stuck with it.

  Bzzzzt! Wrong!

  The customer is stuck, and they tend to remember vendors who
  leave them stuck. They also tell all their friends about such
  vendors.

 -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

  And folks, that's how Digital got to where it is today.

                                   Atlant

    I know I am late jumping into this conversation but I just did a quick
    check for the product in the Chief Engineers Office database and it is
    still listed as a low volume custom build product. Makes me want to
    question the statement about it being declared "Obsolete" even if it
    has been taken out of the price book.
    
    
    Ron 

Ron,

At the risk of being told that I am anti-customer (I'm not - I'm in a SUPPORT
organization)......

The "Dick Best List" is incorrect.  As it relates to almost any OLD CI product,
it has not been updated in years.  In this particular case, the CISCE contains
at least two components which have not been manufactured by ANYONE in the
electronics industry for at least three years (and probably even longer than
that).

The key points in this discussion are these:

   1).	We PRESUME (or maybe someone knows beyond doubt) that there are
	no CISCE amplifier boxes in MCS inventory as spare parts.  God
	help the poor MCS Engineer if one breaks !!!  The good news is
	that the MTBF is much greater than mine.

   2).	There are no parts to build or repair the current design.

   3).	If any person has a compelling need for a CISCE like product,
	either for new installion or to service existing customers,
	then they need to present their case to someone in the appropriate
	business unit to MAKE IT HAPPEN !!! 

/mj