swatt@noc.net.yale.edu (Alan S. Watt) (09/07/90)
>From cisco-request@spot.colorado.edu Fri Sep 7 04:34:15 1990 >Date: Fri 7 Sep 90 00:35:34-PDT >From: William "Chops" Westfield <BILLW@mathom.cisco.com> >Subject: Re: LAT-Telnet Translation over Ethernet Media >To: RAF@cu.nih.gov >Cc: cisco@spot.colorado.edu >In-Reply-To: <9009070317.AA27586@alw.nih.gov> > > >Well, the added speed of the csc/3 processor is sort of irrelevant in >the smaller boxes but the 4 meg of memory (vs 1 meg on the csc/2) come >in handy. Us engineers have this habit of writing lots of code... > >Not only that, but them standards comitees keep making standards use >up lots of memory too. (For example, SPF routing algorthms use more >memory than distance vector algorithms, ISO routes on a per-host >basis instead of a per-net basis, tn3270 takes at least an additional >2k bytes per connection, tcp header compression has to save a bunch >of IP headers for each "interface", and so on.) > >4Meg used to seem like a lot of memory. Now I wish the next generation >of chips would get cheaper faster. > >Bill Westfield >cisco Systems. >------- > Okay, I'm going to air a gripe about cisco in this respect. I'm being told I should upgrade to the CSC/3 processor card because it has more memory, the implication is that soon the 1 megabyte in the older cards won't be enough. At an upgrade price of $4500, I am paying about $1500 per megabyte for these three extra megabytes. This is bad enough, but if this truly is the "decade of ISO" and ISO routes on a per-host basis (didn't they learn anything from DECNET?), then how soon will it be before 4 megabytes isn't enough? And what will I have to do then, upgrade to the CSC/4? I think cisco should immediately re-engineer the CSC/3 board to allow addition of memory (perhaps even on a separate card connected by ribbon cable) *without* having to replace the processor card. You can call this the CSC/3+ or CSC/4 or whatever, but you should do this immediately and offer to exchange existing CSC/3 cards for the new ones on the most generous terms you can afford. Look at it this way: if you don't provide for adding memory, you are going to have to deal with me and a bunch of other customers demanding special software builds with only the protocols we need to save space. If you do solve the problem, you will deal with the same customers by selling them expansion memory boards. Which business would you rather be in? - Alan S. Watt High Speed Networking, Yale University Computing and Information Systems Box 2112 Yale Station New Haven, CT 06520-2112 (203) 432-6600 X394 Watt-Alan@Yale.Edu Disclaimer: "Make Love, Not War -- Be Prepared For Both" - Edelman's Sporting Goods [and Marital Aids?]
hedrick@athos.rutgers.edu (Charles Hedrick) (09/09/90)
I've already commented on this before to people within cisco, but it's probably worth adding my voice to the public comments. A number of our departments recently finished upgrading from CSC1's to CSC2's. It is too soon to be forcing them into another upgrade, particularly to a system that is not expandable. I would strongly suggest the use of SIMM's or something similar.
schoff@psi.com (Martin Lee Schoffstall) (09/10/90)
Alan, An alternative solution is a memory upgrade for the csc/2 to 4Mbytes. Marty ----------- >From cisco-request@spot.colorado.edu Fri Sep 7 04:34:15 1990 >Date: Fri 7 Sep 90 00:35:34-PDT >From: William "Chops" Westfield <BILLW@mathom.cisco.com> >Subject: Re: LAT-Telnet Translation over Ethernet Media >To: RAF@cu.nih.gov >Cc: cisco@spot.colorado.edu >In-Reply-To: <9009070317.AA27586@alw.nih.gov> > > >Well, the added speed of the csc/3 processor is sort of irrelevant in >the smaller boxes but the 4 meg of memory (vs 1 meg on the csc/2) come >in handy. Us engineers have this habit of writing lots of code... > >Not only that, but them standards comitees keep making standards use >up lots of memory too. (For example, SPF routing algorthms use more >memory than distance vector algorithms, ISO routes on a per-host >basis instead of a per-net basis, tn3270 takes at least an additional >2k bytes per connection, tcp header compression has to save a bunch >of IP headers for each "interface", and so on.) > >4Meg used to seem like a lot of memory. Now I wish the next generation >of chips would get cheaper faster. > >Bill Westfield >cisco Systems. >------- > Okay, I'm going to air a gripe about cisco in this respect. I'm being told I should upgrade to the CSC/3 processor card because it has more memory, the implication is that soon the 1 megabyte in the older cards won't be enough. At an upgrade price of $4500, I am paying about $1500 per megabyte for these three extra megabytes. This is bad enough, but if this truly is the "decade of ISO" and ISO routes on a per-host basis (didn't they learn anything from DECNET?), then how soon will it be before 4 megabytes isn't enough? And what will I have to do then, upgrade to the CSC/4? I think cisco should immediately re-engineer the CSC/3 board to allow addition of memory (perhaps even on a separate card connected by ribbon cable) *without* having to replace the processor card. You can call this the CSC/3+ or CSC/4 or whatever, but you should do this immediately and offer to exchange existing CSC/3 cards for the new ones on the most generous terms you can afford. Look at it this way: if you don't provide for adding memory, you are going to have to deal with me and a bunch of other customers demanding special software builds with only the protocols we need to save space. If you do solve the problem, you will deal with the same customers by selling them expansion memory boards. Which business would you rather be in? - Alan S. Watt High Speed Networking, Yale University Computing and Information Systems Box 2112 Yale Station New Haven, CT 06520-2112 (203) 432-6600 X394 Watt-Alan@Yale.Edu Disclaimer: "Make Love, Not War -- Be Prepared For Both" - Edelman's Sporting Goods [and Marital Aids?]
bruce@ccavax.camb.com (09/12/90)
In article <25834@boulder.Colorado.EDU>, swatt@noc.net.yale.edu (Alan S. Watt) writes: > > I think cisco should immediately re-engineer the CSC/3 board to allow > addition of memory (perhaps even on a separate card connected by > ribbon cable) *without* having to replace the processor card. You > can call this the CSC/3+ or CSC/4 or whatever, but you should do this > immediately and offer to exchange existing CSC/3 cards for the new Wonderful idea, and while you are at it, INCLUDE provision for /2 owners (especially those who bought so recently they COULD have ordered /3s - had they been told of their planned arrival...) to not get so ripped off in the upgrade. ANY spare card space should be socketed for some user friendly memory insertion. Catering to 1 and 4 meg SIMMs would currently make a lot of sense. I can think of 2 different basically equivalent X terminals where one chose memory upgrades to be their proprietary cards at about $600 per meg, and the other vendor simply socketed for the same 8 bit SIMMs one stocks in the stationary cabinet for the MAC users. Guess which X terminal sells! Users really appreciate vendors that let them take advantage of commodity priced and available components when building up what is already an EXPENSIVE piece of hardware. Sell us what needs your expertise, but don't try 'DECing' us for the rest.