Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Linux Software

Networking Companies - Eh on Linux 52

netstat sent us the story about the big networking companies and Linux. Much the same story as elsewhere, they already support NT and Solaris, and don't see the user base for Linux there yet. Much of the comments are candids taken from Networld Interop. As you would expect, most of them want to see more users before rolling out support.
This discussion has been archived. No new comments can be posted.

Networking Companies - Eh on Linux

Comments Filter:
  • by Anonymous Coward
    Uhm... Why would that be a problem, unless they believe they'll get all of humanity as subscribers?

    I assume you mean 16 bit.

    Btw., if they're setting up something for that many users, they should be using multiple boxes, and do mail routing on a front system that doesn't have user data. 16 bit or 32 bit uid's isn't any problem.

    Or they could use for instance qmail, and a cdb database to map users to directories, and a checkpassword program that use radius etc. for authentication. Works smoothly, and doesn't require that you add a login account for each user.

  • Take a look at this press release [cai.com]. It's from March, about how CA will distribute a million free copies of Unicenter TNG for Linux together with RedHat.

    I still believe that press release was the reason why one of the companies I do contract work for finally saw the light, and asked me to help them migrate their dedicated network monitoring boxes to Linux... And since several large telcos use those boxes, Linux will go unnoticed in the backdoor at the data centers of some pretty big companies :-)

  • by Anonymous Coward on Friday May 14, 1999 @03:55AM (#1892743)
    I gave last week a talk at SD "Building Web Apps using Linux". It was talk about Apache, PHP and Linux. And as such I was expecting a fairly large crowd. What happened? The crowd never materialized. There was an ok sized crowd Why? The reason is because it takes time. .

    I realized this with another talk I give. This talk I have given for two years. Only this conference is there actually interest.

    The conference folks and I are not fretting because there is interest. People want to see, but it takes time. Rome was not built in a day. Do you remember Windows? Windows took 5 years to catch on after the first commerical adoption. Linux while being around for a long time has only really caught the eye of the public for about two years. That is still a baby. Give it another three years and we shall see.

    I surely hope that the Linux people can rise above this "negative" comments and continue. Remember the success of LINUX is not something companies want. Consider the following.

    Corporate companies believe in a pyramid information approach. The top of the pyramid is the company. They are the top because they create the information. Those that want to share in the pyramid fight to get near the top. However, since the company is at the top they can select who gets the near the top. They can control who becomes their friends and enemies. The size of the pyramid depends on the amount of users who want the information. And the companies compare against each other on the size of the pyramid.

    Enter LINUX. There is no pyramid. Information is not hidden, but shared. There is no one specific entity controlling who gets the information. And as such it cannot be controlled who gets and does not get the information. This scares companies to death. You have just wiped out their business model. In an information age money is made, by those that have hidden information.

    But I think this model is catching on. Especially among the young ideal programmers. Me I am a bit older and I have a problem with the "Open" nature of LINUX. But I am trying to adapt. And so are many other people. Be patient we will get there.

    Christian Gross
  • Instead of trying to market to a user base that isn't there, market to the hobbyists, I always say. If Linux isn't a corporate phenomenum, it's the student movement of the 90's.
  • Mabye Cook's comments are indicitive of why Bay went down the tubes. I didn't replace an old Bay switch with $70k worth of 3com stuff for no reason. Bay just didn't have what I was looking for.

    I'm really worried that the big network companies have incompatible implementations of too many things (like routing protocols, VPNs, VOIP, etc.). I've got [mostly] unmanaged Cisco, 3com and Nortel gear because I bought the best gear for the job, rather than sticking with one brand. What happens when I want to do IP telephony in a managed network?

    I think the networking companies are even more afraid of standards than Microsoft. I hope they get with the program. I really don't want to be forced to buy compatible network equipment exclusively from startup companies making equipment with Linux. I want to be able to use Linux where it makes the most sense and some other vendor's equipment where it makes the most sense.

    -dave

    P.S. What's Cooks email address?
  • uid/gids are 16 bits (at least on x86 linux)
  • Well the ISP I run is _fully_ linux.

    12 assorted x86 machines, just an ascend router and a tigris remote access server.

    We don't have an NT ASP box yet, and probably never will I have anything to do with it.

    Last years Australian ISP survey indicated that over half the ISPs in Oz use linux on the majority of their machines, only 19% used NT exclusively, most that had NT machines only used them for ASP serving for virtual clients, some RADIUS servers too (the reason escapes me).

    Sorry I can't post the pdf with that in it, as it was given free only to participants of the survey and I am not allowed to leak the info. (oops!)

  • It seems we're constantly caught in a the "Catch 22". PHB's won't turn to Linux until they see more support, and vendors won't support Linux until they see more users.

    I know there are lots of vendors who *do* support Linux (24/7 phone support etc, and hardware manufacturers like Mylex etc.), but more are definately needed.

    Of course, I guess what I just said is obvious, so nevermind. ;-)

    - Rev. Randy

  • I worked for a couple of years writing OS/2 device drivers and learned, in the process, how vendors decide how to spend their driver development dollars.

    First of all, a lot of driver development was funded directly by IBM. The Linux equivalent are the user-written drivers (thanks, Don!) out there now.

    But not all the drivers were written by IBM. In fact, you can still find OS/2 drivers on fairly recent products. What motivated vendors to write OS/2 drivers and why won't they write Linux drivers for a similarly-sized market?

    The important answer is that hardware companies don't see the market as a collection of end users, but as a collection of buyers: small, medium and large. The large buyers dictate what the products look like.

    If 20000 people write and say "Give us Linux," that is considerably less important than one person calling up and saying,"I want 20000 machines with Windows 98 on 19995 of them and Linux on 5 of them." If you look at the cost per sale and the cost of custom configurations, that volume buyer will get her Linux, but the 20000 individuals will get the run-around. In other words, the "Market Size" argument is bullshit; it's a sales volume issue, but more complex than "how many Linux boxen." In this example, a 20000-unit sale hinged on only 5 corporate Linux users, just as it once hinged on a small percentage of corporate OS/2 users.

    So, back to the peripheral vendors: Let's say that Dell has just announced they are going to ship Linux (oh wait, they did). Now, they probably did this because a Fortune 500 CIO asked their VP of Sales "what [their] Linux strategy" is. Viola! Now they have one -- not for us, but for a small number of key volume clients.

    But Dell doesn't make their own peripherals. They buy video chips, SCSI chips and so forth. So, having decided to support Linux, they go to their vendors and say,"So, what's your Linux strategy?" Now, even if Dell only sells one machine with Linux on it, it has to run. So are they going to buy millions of SCSI chips from the vendor who doesn't have a Linux driver? Of course not. So... do you want to be the one chip vendor who doesn't have a Linux driver? Duh, no.

    Once upon a time, OS/2 had that "necessary bullet item" status. Vendors shelled out thousands of dollars to consultants (like me) to write the obligatory OS/2 driver. OS/2 flamed out of the marketplace and nowadays most vendors happily ignore the OS/2 market. Linux is hitting that "hot" phase right about now. Unlike OS/2, Linux doesn't have a parent company to abandon it, so it's unlikely to decrease in popularity.

    Bottom line: expect and exponential growth in hardware support, particularly server stuff.

  • Nah. In 10 years, the young'uns will be cursing us for using that outdated Linux, whilst these new massively parallel minicomputers that have 16mil CPU's and 64tb of RAM and run Whiz-Bang OS are MUCH better. OK, so they don't have all that much industry support yet, but look what it can do!
    8-)
  • I just sent this email to the Bay-ISP mailing list (bay-isp-request@bit.net.au, you guys know the drill..) comments?

    ---- Begin Email ----
    I found this at http://www.data.com/story/TWB19990513S0030

    "To add this [Linux] operating system along with the ones we already support
    [which include Sun Microsystems Inc.'s Solaris and Microsoft Corp.'s
    Windows NT], I'd need a lot of users banging on my door," said Christopher
    Cook, product manager for Optivity NETarchitect at Nortel Networks Corp.,
    Billerica, Mass.

    Hello? Nortel? *knock knock*? How hard do I have to bang?

    I'm sure that -someone- from Nortel knows this person, and can take a clue
    stick, go and hit him -hard- with it, amd then hand him this email.

    How hard is it to port from Solaris to Linux? Take the source. Copy it
    across.
    type 'make'. Put a sticker on it saying 'This is only a supported product
    when
    used in conjunction with RedHat 6.0' or Debian, or SuSE or some reasonably
    popular installation, basically whatever platform you typed 'make' on.
    Speak
    to your engineers. They probably -allready- have a linux version working,
    that
    you don't know about (this is what happend at Oracle. Now their highest
    volume
    in database sales is Oracle8 on Linux, seconded by Oracle 8i on Sparc. If
    they
    weren't selling a linux version, people would be using postgres, mysql,
    msql,
    or any number of other lesser known, but available databases.)

    Perhaps you should think about it this way. Only last week, I spoke to my
    sales guy (Hi Grahame!) about a decent network management solution. As I
    -am-, basically, a bay shop, I started thinking about Optivity. Only having
    a
    few (rather decrepit) Sun's lying around, and all my grunt in Intel
    hardware,
    I started having a look at the supported platforms.. Uh. No Linux there. Ah
    well.. You miss out on a sale.

    For me to buy a 'decent' sun box (eg, ultra5 or something) I'm looking at
    what, $6k these days for a lightly spec'ed ultrasparc? Then another $5k for
    Optivity. Total outlay, $11k. Now, instead of spending $11k, I could take 2
    weeks off work (costing myself, say, $5k) and cobble together something
    that's
    not as nice, but does what I -need- it to do.

    Yeah, sure, I'd love to have Optivity. It's nice. But I don't wanna go and
    put
    an -unreliable server- as my network management workstation (eg, NT) , and
    then have it freeze with everything green, whilst the network around me
    crumbles into a heap. I don't want to spend more than double what I need to
    on
    buying an ultrasparc to manage my network.

    All -you- guys have to do is release Optivity on Linux. People will buy it,
    and suddenly you'll be the good guys. And it's -real- rare for a geek to
    consider a telco a good guy 8-)

    For geek-type discussions on exactly this subject, please feel free to look
    at:

    http://slashdot.org/articles/99/05/14/1258203.sh tml

    I'm sure there'll be -numerous- discussions on there by the time this email
    gets to you 8-)

    One small point of congratulations to Nortel is that I bought a few netgear
    cards the other day. They had 'Linux' as a supported OS on the box. Sure I
    paid a bit more ($7 more, each, than a D-link Tulip based card that was
    sitting right next to it on the shelf - basically exactly the same card) for
    it, but I bought it -because- it said linux on the box.

    --Rob
  • ISP Going to be running all of there mail/web/shell farms under linux. Biggest problem they face is makeing linux 32bit uid/gid aware.
  • The second half of your post refers to "information" and "information" being kept secret.

    What exactly are you referring to?

    The way I see it, companies don't like Linux because they have to be reactive against it, since they have no way of controlling it. Well, they can control it to some extent, but only if they make contributions to improve it and are sincere.

    Could you clarify what you're referring to as control of information? What information?
  • Are you talking about performance of an application on hardware you'd never actually use for those things? Who uses a quad xeon to serve static pages? Why didn't they test Linux using a web server that's optimized for static pages?

    Also, there was an article in http://www.linuxtoday.com [linuxtoday.com] that exposes [linuxtoday.com] the fact that Samba under Linux actually performed better than Windows NT on the PCWeek tests when using NT clients than NT did. This is something that everyone should know.

    The PCWeek numbers were also much higher than the Mindcraft numbers, attacking the credibility of Mindcraft even more.

    Another funny thing, if anyone follows http://lwn.net [lwn.net], they had a story [lwn.net] that mentioned why Apache may have done so badly, and how we can modify the scheduler to make the problem go away.

    If anything, we should thank MS for the Mindcraft tests. It may end up making Linux better in the end.

    I don't think Linux did as bad as people think, and I'm sure that static web serving performance or Samba performance is irrelevant in a router.
  • Agreed; the general feel I got was that they were watching linux, but didn't see the market there at the moment. They are in the business of making money, and they obviously don't feel that the payoff is worthwhile at this time. However, some companies might miss the boat if they suddenly realise that linux is bigger than they thought and they have a 3 month development cycle to get up to speed on linux. If this is the case, companies like Corel, Oracle and others could really start cleaning up. Only time will tell.
    --
  • Uhh.. Lucent's software supports nearly any OS. It's all in Java. Linux is not a special case.
  • Uhh.. half the time, I had a devil of a time getting perfectly "normal" source (meaning, it would compile fine under Solaris, FreeBSD, NetBSD) to compile correctly under RedHat or Debian. The opposite is the same; people who code using those damned Linux-specific header files irk me. Couldn't you just stick to POSIX please?
  • by Cato ( 8296 ) on Friday May 14, 1999 @07:22AM (#1892759)
    HP's OpenView division is porting a number of network management products to Linux - Firehunter is already there and others are coming.

    WatchGuard is a neat firewall device whose key feature is a team of security people who remotely update the rules and software as new exploits are discovered - all Linux based.

    Interestingly, Cisco is going to bundle Apache with its NT-based web-based network management tools, apparently because it's too hard to set up the various flavours of MS IIS correctly.

    Product support will follow Linux market share - the queso OS surveys make it quite clear that Linux is the dominant Web host with 30+ per cent market share, followed by Windows then Solaris - so it's likely that management tools focusing on web hosts will migrate to Linux quite quickly.

    On a personal note... if there's anyone out there who wants bandwidth management / QoS support for Linux web hosting, please email me at rdonkin@orchestream.com as well as commenting here - I'm curious as to how much demand is out there, ideally from large enterprises or service providers. (My company, Orchestream, makes policy-based network management tools for QoS/CoS, and the Linux-DiffServ stuff looks like a very capable platform for this - I think web hosts will be the primary market for any Linux version.)

    The key, as always, is to find the right person in a company (typically a product manager or marketing manager) and contact them with your request for Linux support, including details of the number of users, number of Linux hosts, so they can start salivating over the order size :)
  • "[Linux is] a 'techie' technology right now," said Cam Cullen.

    Oh, for crying out loud, networking is a 'techie' technology. How many end users administer networks? 3Com has some notoriously good engineers, but bigosh, their marketing people are just as thick as those at any other company.

  • Like I said...when today's users reach decission levels, linux not only will have support, it'll control the world...it just take time.
    Our future sucess is guaranteed as long as companies "feel" the need to support linux, and that feeling is only efective in their pockets.
  • Wrong, we are the kind of idiot that would rewrite the entire OS to adapt the new needs, Whiz-Bang OS will be Linux 18.3.17. Because the freedom enviroment in Linux development is what attracts new developers more and more, and because almos every nerd I know, who 4 years ago was writing windoze shareware is now writing linux opensourced apps, in the same free time (free time...that why they call us nerds...;)

    I'm 24, I'm young, and I'm making good money and all thanks to linux.
  • This guy is waiting for more requests on linux???
    God, he's missing the wave!
    I know many companies that without knowing it, are using linux every day. And they do it because after asking solutions to companies like them, they get scared of their prices and run to smaller companies who solve their problems for 1/10 of nortel prices...
    I saw linux boxes used in tiny rooms at big buildings roofs routing ip, and when I asked why I got:
    "We were tired of going up and down the elevator 20 floors twice a day with NT, so we put Linux and now we do it once a month". But nobody in the company knoes except "techies".
    "Techies" will exist always, so why big companies lie to their customers. They won't have a large network running with no techies. If they want it, they have to hire us, and pay good money.
    I say: "Wait 5-10 years, let our generation reach managment levels in organizations and let's see how we run everything from desktops to critical mission apps in Linux". Remeber we are the GNU generation...
  • Whew.... You're a little lost. Companies think they make money because they control:

    Source code: Otherwise they can't sell software.

    Hardware specs: Otherwise (they believe) other people will design clones.

    Passwords: all attempts to sell information off the net rely either on passwords or moral exhortations not to distribute copies.

    Know-how: This, together with source code, enables companies to make $1E8's from customer support. If it were all available for the taking (as with the Linux HOWTOs), you'd never pay for a call. (Redhat survives, of course, but only as a convenience, not a necessity.)

    Patents: This one's pretty obvious.

    Demographics: Spammers of the net, mail, and telephones sell each other information about you. Open source doesn't address this one, but it's part of the mindset.

    Want more examples? I believe a free information society would look more radically different than a society in which physical goods cost nothing.

    You have a good point: companies *do* have to be reactive to Linux unless they "give away" their knowledge to it. But Linux forces companies to be reactive because they don't control its information, and they can't buy out anyone who does.

    AOL needn't be reactive to Netscape anymore because they can buy Netscape and hence have sole control of its information. Mozilla, on the other hand, can do whatever it wants, and AOL probably hates that.

    Jesse.
  • Netgear cards have been shipping with a linux driver for several months now. They include the source for a modified tulip driver so that people who have older distros can have the card working out of the box.

    Even better, the Netgear cards actually display that liunx is supported on the outside of the box.

    I'd say its just a matter of months before 3com follows suit on their cards... and I would bet that most router/switch management packages are in the process of being ported to linux (if they're not already web-based), no matter what the company marketing people think or say.

  • You make an extremely good point, especially with regard to "enterprise" products such as the article mentioned. While the user base might be able to bug Creative or 3dfx into releasing Linux drivers, it's going to take big corporate orders to get the enterprise products on Linux.

    (PS - This must be happening though, if you look at Oracle or DB2 for Linux.)
    --
  • All the tech people I know run Linux and prefer to use it.

    You really should get out and try the real world sometime, it's a pretty nice place.

    Cheers,
    ZicoKnows@hotmail.com
    Slashdot Realist

  • I don't think thats so uncommon, we already do this at the ISP I work for (all linux except a NT box for sites with .asp script, and a fileserver (which works like crap btw)

  • by BobBoring ( 18422 ) on Friday May 14, 1999 @03:42AM (#1892769) Homepage
    Linux users in general don't call for support. Especially slackers like me who are too lazy to E-mail in a bug report. So even though we hack around problems and make stuff work, _we_ never ring the right bells with vendors and PHB's. My experience has been whatever work arounds for the other unixen are there are similar work arounds for linux. We let each other know how to do it but we never spam the ppl who report to the ppl making the _business_ decisions. The moral of the story is call frequently with real issues and you will get a better response from the 'other' minded ones.
  • Actually, to throw a large person, you don't have to pull very hard at all. I've seen a 45kg (~100lbs) woman throw a 95kg (~210) man.

  • From the article:

    "[Linux is] a 'techie' technology right now," said Cam Cullen, product marketing manager of large enterprise served marketing at 3Com [...] "When the big applications vendors say they'll support it, then we'll support it. And believe me, we're tracking what they do."

    Huh? Oracle? Word Perfect? Oh, I'm sorry, by "big applications vendors" he must have meant Microsoft. Well, pardon the heck out of me. I guess we can write 3Com off....

    No, I'm not Microsoft bashing. I'm PHB-bashing.

    --
    10. Thou shalt forever forswear and abjure the vile notion that All The World's A VAX.
    -- Henry Spencer (was henry@utzoo.utoronto.edu), the Ten Commandments for Hackers
  • Well, to some people it is, but I cannot speak for them.

    To me, it is the operating system I use. I do not use Solaris (or SCO UNIX, or NT, or anything else (Well, I boot Win95 to play games. So sue me.)).

    It is really simple. If I work for company XYZ, and I have 50 Linux boxes deployed, and I want network cards for them, I am going to go to a company that supports Linux. If Nortel does not support Linux, they lose a sale.

    As for what Solaris and Linux have or do not have, well, Solaris is generally run on more high-end hardware, and scales like a dream. Linux is cheaper and comes with source code.
  • Is Linux some kind of religion ? They use Solaris already. What else do you need? What Linux can do that Solaris can't ?
  • by RiverRat ( 31033 ) on Friday May 14, 1999 @03:33AM (#1892774)
    Even with an expontential growth curve, big numbers don't happen overnight. Note that they said they are watching. As ERP and other big applications happen, the support will come.
    My martial arts instructor said to throw a large person, you have to pull very hard and for a long time. Big companies take time to be moved. Just keep pushing and don't let up. It will happen no other way.
  • >Linux users in general don't call for support.

    Actually, we do call the companys for support, but in a roundabout way.

    If a card doesn't work for a windows user, they ask their retailer, search on the suppliers web site for the latest drivers, etc, and hence they get noticed by the suppliers.

    Linux users however, when a card doesn't work, check out kernel.org, or post to a newsgroup asking for info. Chances are there is work in progress on the card, and they download a updated kernel and away they go, unnoticed by the suppliers.

    But where do the kernel developers go for support or information on the card? They have to get the card information from somewhere, be that released specifications, trial and error, or whaever. But the best way is to ask the company involved for specs so they can write a efficent and correct driver.

    In both cases, people are asking the company for information, but the difference is that in the first case, its 100,000 people asking for a win9x driver, and in the latter it's say, half a dozen developers asking for technical specs.

    This impacts their thinking in two ways
    a) Many more people are asking *directly* for 9x support
    b) the Linux people are asking for very detailed info, which companys are still reluctant to release.

    We have to convince the people at the top of two things

    a) The linux device driver distribution is more indirect than the windows one, but the amount of *end users* would be comparable.
    b) If they release the specifications in plenty of time, they won't have to "support" linux - we'll have already done it!


    --
  • Uhh.. Lucent's software supports nearly any OS. It's all in Java. Linux is not a special case While Lucent has a Java version of some of their software, their Radius supports Linux natively. They have two compiled versions, one for Redhat and one for Slackware.
  • This is one of the reasons I am using a Portmaster RAS instead of a Total Control. 3com's Radius software will only run on NT or Solarius. Lucent's will run on Linux. I made sure the rep was aware that it was one of the factors in my decision.

    This is more effective in the long run than sending email asking them to port. With the email, they will assume that you will still use their software without it being ported. With a lost sale, there is no misunderstanding.

    I also make it a point of asking salemen at retail stores if they sell Linux software, especially if I know that they don't. Then I get great satisfaction of saying "OK" and walking out of the store.

Things are not as simple as they seems at first. - Edward Thorp

Working...