phil@amdcad.UUCP (Phil Ngai) (08/01/86)
In article <838@usl.UUCP> elg@usl.UUCP (Eric Lee Green) writes: >In article <58@rpics.uucp> schoff@rpics.UUCP writes: >>People interested in CISCO should investigate the COMPANY. Technically >>I see no problems, but it truely is a garage-shop operation, how >>many full time employees do they have? (< 5). > >Wow. This guy must not buy any computer besides IBM, because "Nobody >wants to buy anything from some LITTLE company, right?". It is >thinking such as this that stifles technical innovation.... you build >a better mousetrap, then nobody wants to buy it because you're not as >big as Imperial Business Machines. > >In other words, I find it a disgusting attitude... >-- Computing from the Bayous, -- > Eric Green {akgua,ut-sally}!usl!elg > (Snail Mail P.O. Box 92191, Lafayette, LA 70509) I think information on the size of Cisco Systems is relevant and here's why. When we install a service, people start using it and depending on it to get their job done. Interruptions in service are pretty serious because our customers depend on us to meet our commitments just as we depend on our vendors to meet their commitments. Unfortunately there is more to a product than great engineering. There is manufacturing and support. All the icky boring stuff that customers seem to demand in return for their money. We had a very bad experience with one of our vendors. We had installed about forty pieces of their networking equipment serving about 400 users when their equipment started failing. These were the worst kinds of failures: in the field, intermittent, weeks or months after the installation was completed. Our vendor recognized the seriousness of this problem and started working on it. Eventually they isolated it to a chip vendor which "from Jan 83 to Jun 84 delivered product which exhibited separations or cracking of the metallization after about 500 hours of operation". This was a big disaster for us and for our vendor. For 18 months our vendor had been shipping booby trapped product. Of course, they went back and yanked out all the bad chips, which must have cost them greatly. It also cost us in downtime while finding the problem and then while cleaning out the system. We're still not finished. How many dollars and how many manhours? I estimate our vendor had to correct the problem in thousands of boxes. Assigning a cost of $10 to $50 per box, we're talking as much as $100,000. As for us, we had about a dozen failures. Let's assume 10 engineers unable to work for a day each time a failure occurs. 12 * 10 * 8 * $40 = $38,400. (no we don't pay engineers $40/hour, that's the burdened cost) And that's just the cost in payroll terms, the cost in lost sales is harder to estimate but surely larger. When a company reaches a certain size they have enough experience to know things like the fact that buying from the lowest bidder is not enough. Companies like HP have armies of reliability and qualification engineers to prevent just this kind of disaster. HP, for example, doesn't buy from just anyone who prints up a catalog. You have to qualify to become a vendor for HP. This involves testing hundreds of chips for weeks at a time, among other things. This extra engineering is one ingredient that companies like mine look for from their vendor, because the cost of failures are so high, and one ingredient that a small company like Cisco is less likely to know how to put in. I don't know much about Cisco, they may be a totally together company but the odds are against it and Marty's comment about their size is relevant. Robert, just how well do you know the people at Cisco? Perhaps you can address the issues I have raised in this article. To say >thinking such as this that stifles technical innovation.... you build >a better mousetrap, then nobody wants to buy it because you're not as >big as Imperial Business Machines. shows a lack of knowledge about why companies buy products. Your mousetrap may be more advanced technically but if you can't service it like a bigger company can then I will have to consider the extra (hidden) costs associated with such a product. -- Classical music audiences are like ivory soap: 99 44/100 pure (white). Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com
elg@usl.UUCP (Eric Lee Green) (08/04/86)
In article <12538@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes: >In article <838@usl.UUCP> elg@usl.UUCP (Eric Lee Green) writes: >>In article <58@rpics.uucp> schoff@rpics.UUCP writes: >>>People interested in CISCO should investigate the COMPANY. Technically >>>I see no problems, but it truely is a garage-shop operation, how >>>many full time employees do they have? (< 5). >> >>Wow. This guy must not buy any computer besides IBM, because "Nobody >>wants to buy anything from some LITTLE company, right?". It is >>thinking such as this that stifles technical innovation.... you build >>a better mousetrap, then nobody wants to buy it because you're not as >>big as Imperial Business Machines. >> >>In other words, I find it a disgusting attitude... [Relates horror story about company that bought bad chips and is still ferretting them out, because of poor quality assurance program. Then brings up the example of HP, one of the few companies with a very strict quality assurance program -- and high prices to match! ] >To say > >>thinking such as this that stifles technical innovation.... you build >>a better mousetrap, then nobody wants to buy it because you're not as >>big as Imperial Business Machines. > >shows a lack of knowledge about why companies buy products. Your >mousetrap may be more advanced technically but if you can't service it >like a bigger company can then I will have to consider the extra >(hidden) costs associated with such a product. Yes, I would investigate Cisco, look at their quality control, service, etc. However, I would do the same with any manufacturer, regardless of size. Remember DEC's failing RA80s? and how DEC continues to deny that there's any problem? It is not the size of the company, but rather the service the company provides, which is at issue. Some small startups never make the transition needed between small sales/small volume and large sales/large volume, i.e. setting up the management infrastructure necessary. However, there's many examples of companies that have a good product, that have the funds, that have the service personel, who don't make it because they aren't "IBM compatible" (if you're talking PCs) or "It's not a VAX" (if you're talking super-minis) etc. USL has three Pyramid 90x's, bought when Pyramid Technologies was a small startup. They have provided good service in their intended duty. If we had bought equivalent Vaxen of that era, they would have been severely overloaded and slower than mollasses under a load of about 25 students average per computer. Should we have bought three Vaxen because "Pyramid is just a little startup that could go out of business tommorrow"? Perhaps I'm confusing a university environment with a commercial environment... when something breaks here, generally the answer is to look around among us for an expert to fix it, rather than call in the serviceman and wait for him to fly in from Massachusetts or California. Do corporations lack this large body of talent? If so, why? And if so, how do American corporations intend to compete with Japanese corporations, which place great emphasis upon hiring talent and conducting large amounts of research? Your average Japanese auto executive started out assembling cars on the floor. If his car breaks, he can fix it. The average American auto executive started out in Harvard Business School. If his car breaks, he must call a mechanic. Maybe that's why the Japanese build better cars! In other words, I still find that people overlooking a company merely because of its size is pretty nasty... but then again, as a proud co-partner of a small telecommunications firm running out of a back bedroom, I'd naturally think that :-). (our primary strategy is to provide both a better program and better service than our big competitor, which is sort of cocky and snaps at the users). -- -- Computing from the Bayous, -- Eric Green {akgua,ut-sally}!usl!elg (Snail Mail P.O. Box 92191, Lafayette, LA 70509)
henry@utzoo.UUCP (Henry Spencer) (08/06/86)
> Yes, I would investigate Cisco, look at their quality control, > service, etc. However, I would do the same with any manufacturer, > regardless of size. Remember DEC's failing RA80s? and how DEC > continues to deny that there's any problem?... Or ask that paragon of service and support, IBM, about the problems with the PC/AT hard disks. "What problems?" -- EDEC: Stupidly non-standard brain-damaged incompatible Henry Spencer @ U of Toronto Zoology proprietary protocol used. {allegra,ihnp4,decvax,pyramid}!utzoo!henry
michaels@hplabsb.UUCP (Robert Michaels) (08/12/86)
In article <7020@utzoo.UUCP>, henry@utzoo.UUCP (Henry Spencer) writes: > > Yes, I would investigate Cisco, look at their quality control, > > service, etc. However, I would do the same with any manufacturer, > > regardless of size. Remember DEC's failing RA80s? and how DEC > > continues to deny that there's any problem?... > Here is some info about the design and construction of the csico hardware: The box is mulibus unit with a 9 slot card cage. All the sheet metal including card cage are cisco designed and built. The power supply and cooling blower are OEM. The card cage contains a cpu card, ethernet card and 2 (16 line) terminal interface cards. All but the ethernet interface card are desgined and built by cisco. The ethernet card can be either interlan or 3Com. The cpu card and terminal line cards are farily simple straight forward design. As far as I can tell he uses no exotic ic's that couldn't be replaced by a different vendor. The cpu consists of a 68000, 1 Mb of RAM and room for 4 byte wide EPROMS. The terminal line card uses 8 2681s which are fairly well known uarts. I don't know who make the power supply. The blower is some German built unit which apparently can move lots of air. It is quite noisy, but it was designed for use in farily warm environments like wiring closets. It is my opinion (for whatever its worth) that cisco hardware will not have any significant problems. Their service attitude is they will send you a new board if a problem occurs. I think that most corporations and universities can live with that policy. I have heard, although I'm not certain, that cisco has a farily flexible policy regarding source code. I'm not sure what it is but it is supposed to be possible for most customers to obtain a copy given the usual restrictions. - Robert Michaels ( michaels@hplabs ) DISCLAIMER: Please understand that these are my opinions and not that of Hewlett-Packard Corporation.
swb@batcomputer.TN.CORNELL.EDU (Scott Brim) (08/14/86)
In article <3653@hplabsb.UUCP> michaels@hplabsb.UUCP (Robert Michaels) writes: >I have heard, although I'm not certain, that cisco has a farily flexible >policy regarding source code. I'm not sure what it is but it is supposed >to be possible for most customers to obtain a copy given the usual >restrictions. At the end of June cisco told me they did not yet have a policy on source availability, and they would have to talk to a lawyer about the possibility of an escrow agreement, etc. We were talking about gateways, not terminal servers. ------------------------------ Scott W. Brim swb@cu-arpa.cs.cornell.edu Cornell Theory Center {decvax,ihnp4}!cornell!swb 265 Olin Hall bitnet: swb@crnlcs Cornell University 607-255-9392 Ithaca, NY 14853
phil@amdcad.UUCP (Phil Ngai) (08/16/86)
In article <857@usl.UUCP> elg@usl.UUCP (Eric Lee Green) writes: >Some small startups never make the transition needed between small >sales/small volume and large sales/large volume, i.e. setting up the >management infrastructure necessary. However, there's many examples of >companies that have a good product, that have the funds, that have the >service personel, who don't make it because they aren't "IBM >compatible" (if you're talking PCs) or "It's not a VAX" (if you're >talking super-minis) etc. Well, I happen to believe the marketplace is where a product proves how good it is. If it doesn't sell, then by definition it is not a good product because it does not meet the needs of the customer. If a customer wants to run Lotus and buys a PC instead of a Mac then the Mac maker blew it in his market research. >USL has three Pyramid 90x's, bought when Pyramid Technologies was a >small startup. They have provided good service in their intended duty. >If we had bought equivalent Vaxen of that era, they would have been >severely overloaded and slower than mollasses under a load of about 25 >students average per computer. Should we have bought three Vaxen >because "Pyramid is just a little startup that could go out of >business tommorrow"? It's nice that your decision worked out for you but if Pyramid had failed I think you wouldn't have been so happy about it. >Perhaps I'm confusing a university environment with a commercial >environment... when something breaks here, generally the answer is to >look around among us for an expert to fix it, rather than call in the >serviceman and wait for him to fly in from Massachusetts or >California. Do corporations lack this large body of talent? First of all, we think the manufacturer should design it to reduce the chance of it failing. Quality should be designed and manufactured in, and I expect small companies to be less likely to have the expertise to do this properly. Not that they can't, but that they are less likely and therefore it is a valid question as to how large a vendor is. As for fixing the machines we have better things to do than play field service drone. Maybe if we had a large pool of slave labor (students) we might do what you do. But if you had to pay for your manpower perhaps you would do what we do. >And if so, how do American corporations intend to compete with >Japanese corporations, which place great emphasis upon hiring talent >and conducting large amounts of research? Your average Japanese auto >executive started out assembling cars on the floor. If his car breaks, >he can fix it. The average American auto executive started out in >Harvard Business School. If his car breaks, he must call a mechanic. >Maybe that's why the Japanese build better cars! Why should a semiconductor designer know how to fix a car? Or a computer? >In other words, I still find that people overlooking a company merely >because of its size is pretty nasty... but then again, as a proud >co-partner of a small telecommunications firm running out of a back >bedroom, I'd naturally think that :-). (our primary strategy is to >provide both a better program and better service than our big >competitor, which is sort of cocky and snaps at the users). If you can't figure out what (rightful) disadvantages you have as a small company, you may never become a big company. Stop complaining the world isn't the way you want it and try to understand how it really works. I'd like to point out that we seem to be talking about two different levels, software and hardware. It takes a lot of work to build high quality hardware. Parts qualification is one thing I'm thinking of. You have to test a large sample of each chip from each vendor. How can a small company afford to buy a thousand 68020s just to bake them in an oven and see how long they work? Quality in software is a little different. If you burn enough midnight oil you can probably produce as good a product as a big company. Although I don't know if your manuals will be as good. Or if you will offer training and system engineers to help customers. Maybe you can. But that's the kind of thing big customers (you know, the kind with lots of money) want. -- Rain follows the plow. Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com
kaufman@Shasta.STANFORD.EDU (Marc Kaufman) (08/16/86)
In article <12721@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes:
.>... Parts qualification is one thing I'm thinking of.
.>You have to test a large sample of each chip from each vendor. How
.>can a small company afford to buy a thousand 68020s just to bake them
.>in an oven and see how long they work?
Actually, you don't really need to do this. Just buy from the Japanese.
They actually test their parts before they ship them, unlike the
American companies.
phil@amdcad.UUCP (Phil Ngai) (08/17/86)
In article <772@Shasta.STANFORD.EDU> kaufman@Shasta.UUCP (Marc Kaufman) writes: >In article <12721@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes: >>... Parts qualification is one thing I'm thinking of. >>You have to test a large sample of each chip from each vendor. How >>can a small company afford to buy a thousand 68020s just to bake them >>in an oven and see how long they work? > >Actually, you don't really need to do this. Just buy from the Japanese. >They actually test their parts before they ship them, unlike the >American companies. You're wrong, some American companies do test their parts before shipment. Also, I was specifically thinking of problems that only show up after a few hundred hours of operation. -- Rain follows the plow. Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com