crossover and straight thru

Discussion in 'Cisco' started by CREAM, Oct 10, 2010.

  1. CREAM

    CREAM Guest

    hi, im trying to figure out crossover and straight thru. I know the
    similar devices like switch to switch and router to router requires a
    cross over. I know that like switch to a router requires a straight
    thru. i came across a article on the internet, it said that a Nic to a
    router is crossover cable? I might be wrong here but isn't a nic to
    a router a straigt t hru since they are different devices Thanks for
    the help.
     
    CREAM, Oct 10, 2010
    #1
    1. Advertisements

  2. CREAM

    bod43 Guest

    I don't know the official terms but there are two kinds of
    devices. PC, routers, printers, fax machines are
    all one kind. Switches are the other kind.

    Originally for UTP Ethernet it was hubs one way and
    everything else the other way which was simple enough.
    PC to hub was straight through, hub to hub or PC-to-PC
    needsd a crossover.

    Then we got routers and bridges which were not hubs
    and so were in the everything else category.

    Then switches replaced hubs.

    Now of course some switches are really routers
    as well as being switches:)

    A router is normally the same as a PC
    and it needs a crossover to connect to a PC.

    Of course many cisco and other switches now auto
    detect and work with either cable (auto-mdix) and
    as far as I recall GBE (gigabit Ethernet) works with
    either too - always.
     
    bod43, Oct 10, 2010
    #2
    1. Advertisements

  3. In general every-day practice, this isn't something you have to worry
    about at all. Use straight through cables everywhere, and don't worry
    about having a crossover cable anywhere, because most likely any that
    you get will be made incorrectly for GigE speeds and break things for you.

    GigE requires auto detection of the link if its MDI or MDI-X, and you
    should use straight throughs.

    If you are concerned about it on an academic level, then yes, any
    device that doesn't do auto MDI/MDI-X detection, a cross-over is
    required if you hook a switch to a switch, or a NIC to a NIC, they
    need to cross-over to talk to one another. Any point where a switch
    connects to a NIC, then its straight through.

    A router port would be a NIC type connection (unless you have a level3
    switch, or a switch-blade in your router).
     
    Doug McIntyre, Oct 10, 2010
    #3
  4. CREAM

    Sam Wilson Guest

    "Hub" has always been a confusing term; the first UTP-ish ethernet
    things were repeaters.
    Switches replaced repeaters, but they were a direct replacement - no
    wiring changes needed (usually).
    Don't confuse the children! If it's capable of being a switch (hub)
    then it's a switch for this purpose. If it's only a router then it's
    only a router.
    This is true. You can usually connect two reasonably modern PCs
    together with a straight through cable and they'll work - but see below.
    GigE (1000base-T) is specced to work with a straight through cable -
    always! That's because it uses all four pairs of wires in both
    directions simultaneously. Using a crossover cable (of which there are
    at least two sorts, but let's not go there) really ought to confuse it,
    but the standards guys were canny enough to see that coming and allow
    crossover cables. AFAICT you should never need one for GigE (anyone
    care to differ?) and the fact that many PCs have 10/100/1000 ports in
    them these days means that a straight through cable will work for them.
    Needless to say you shouldn't need a crossover to connect two routers or
    two switches with 1000base-T interfaces.

    HTH

    Sam
     
    Sam Wilson, Oct 11, 2010
    #4
  5. Cisco has some old Copper GigE GBICs that don't support auto MDI/MDI-X.

    If you have those in two Cisco switches, then you'll need to construct
    a GigE crossover cable yourself, because whatever you find in the
    stores is most likely to be fubar and wrong.

    They fixed that years ago, but if you have some old old gear, its
    definately a possibility.
     
    Doug McIntyre, Oct 11, 2010
    #5
  6. CREAM

    Rob Guest

    Presumably those don't comply to the widely used 1000BASE-T standard?
    Maybe it is 1000BASE-TX instead?
     
    Rob, Oct 12, 2010
    #6
  7. CREAM

    bod43 Guest

    Ah yes, I *never* use the term "hub" for the reasons
    you mentioned. Except this once it seems:-((

    Thanks for fixing it. Dunces corner for me.
    It seems to me that a cable with all 4 pairs crossed
    will be indistinguishable *by principle* from a straight
    through and so the standards people had no additional
    work to do.

    I am guessing. I don't actually know how it works:)

    If the "end host" and "concentrator" ports are identical
    then the above will be the case. The port can't "know"
    what it is talking to or whether a cross cable is in the
    line or not.

    No additional work from the designers.

    If the ends are not identical and say each signal their
    "preferred" type to the other then of course the case of
    a crossover cable will need to be considered by the
    end devices.

    Most crossovers I encountered have been full 4 pair crossovers
    and I am pretty sure I have used them on GBE ports.

    I will post this to comp.dcom.lans.ethernet (probably the
    first on-topic post for months) too and perhaps
    Mr Seifert will comment on whether either end
    ever becomes aware that there is a crossover there
    or not? It's not important at all now
    since in my experience it all magically just works
    so maybe he will consider it not worth the time?
     
    bod43, Oct 12, 2010
    #7
  8. No, they definately were for 1000Base-T. Its possible that Cisco
    released them before the 802.3ab standard was out (1999?), as I
    remember having the problem around or before that year.

    Anyway, auto MDI/MDI-X is optional in the 1000Base-T standard.

    IEEE 802.3-2008
    40.4.4 Automatic MDI/MDI-X Configuration
    ...
    Implementation of an automatic MDI/MDI-X configuration is
    optional for 1000BASE-T devices. If an automatic configuration
    method is used, it shall comply with the following specifications.
    ...

    I haven't run across anything since then that hasn't supported auto
    MDI/MDI-X though.
     
    Doug McIntyre, Oct 12, 2010
    #8
  9. CREAM

    Rick Jones Guest

    My contribution to terminology based on vague and error-prone
    recollections of "the old days:"

    repeater - two ports - what is seen on one port is always repeated (at
    the physical layer) on the other port

    hub - a repeater with more than two ports. what is seen on one port is
    always repeated on the remaining ports.

    bridge - two ports - what is sean on one port is always forwarded at
    the data link layer on the other port

    learning bridge - a bridge that learns which MACs are on which side
    and makes forwarding decisions based on that

    switch - a bridge with more than two ports.

    learning switch - a switch that learns which MACs are on which port
    and makes forwarding decisions based on that.

    some might replace "learning" with "smart"

    I might yield on learning/smart vs not on terminology and go with "dumb
    bridge" as the one that does not learn and "bridge" as one that does -
    same for switch.

    rick jones
     
    Rick Jones, Oct 12, 2010
    #9
  10. For most 10baseT and 100baseTX ports, transmit and receive
    are on separate pairs, and must be straight or crossed
    as appropriate.

    When I do make crossover cables (not very often), I cross
    the 100baseTX pairs (1,2) with (3,6), and also pairs
    (4,5) with (7,8). I have heard that others don't cross
    the latter.
    Well, for gigabit, which both sends and receives on all
    four pairs, it isn't so hard for it to figure out which
    pair connects to which, and extract the bits accordingly.

    Also, many 100baseTX ports have this ability, presumably
    with extra hardware.
    -- glen
     
    glen herrmannsfeldt, Oct 12, 2010
    #10
  11. CREAM

    Rich Seifert Guest

    No, these two cases *are* distinguishable because the signals on the
    four pairs have different semantics (meanings). The 1000 Mb/s data
    stream is sent as four streams carrying 250 Mb/s each, and then
    recombined at the far end. It is important that the signal be
    recombined in the proper order, so you cannot "mix up" the wire
    pairings indiscriminately.

    Also, Auto-Negotiation (which is mandatory in 1000BASE-T) is performed
    on only two pairs--the same two pairs that are used for 10/100BASE-T.
    This is necessary because it is entirely possible that a 1000 Mb/s
    device at one end is Auto-Negotiating with a 10/100 Mb/s device at the
    other end, with the result of 100 Mb/s operation. So, again, the
    pairing is important.

    Of course, it is easy to figure out whether you have a straight-
    through cable (Auto-Negotiation signals from the other end will show
    up at pins 1/2 on your input) or a crossover cable (Auto-Negotiation
    signals from the other end will show up at pins 3/6 on your input).
    Once determined, you can organize the data signals for proper
    presentation to the four pairs : 1/2, 3/6, 4/5, 7/8 so that they can
    be decoded and recombined at the other end.
    *Some* additional work from the designers, but not a whole lot. Even
    before Gigabit Ethernet, many Ethernet chip manufacturers had
    incorporated "auto-crossover-detect" in their 10/100 Mb/s products,
    because it made life easier for the installers and network
    administrators--no manual setting of MDI/MDI-X switches on devices,
    and it doesn't matter which flavor of cable you use.
    There has indeed been a dearth of on-topic postings to this newsgroup;
    I only check it every week or so now. This is not a statement about
    interest in Ethernet, but a statement about the general demise of
    Usenet.

    --
    Rich Seifert Networks and Communications Consulting
    21885 Bear Creek Way
    (408) 395-5700 Los Gatos, CA 95033
    (408) 228-0803 FAX

    Send replies to: usenet at richseifert dot com
     
    Rich Seifert, Oct 15, 2010
    #11
  12. CREAM

    bod43 Guest

    Thanks Rich illuminating as always.

    Of course usenet has obviously declined and sadly
    no clearly defined replacement is forthcoming. Mostly
    in my view this is due to the more recent development
    of tools that make it trivial for any Joe to knock up their
    own version of any forum. There have been many cases of
    people making usenet posts requesting that users
    divert their attention to competing forums.

    I have no idea where to find the equivalent of
    comp.dcom.lans.ethernet or comp.dcom.sys.cisco
    as they were only a few years ago. When I first
    encountered them (along with the now obsolete
    lans.fddi and lans.token ring newsgroups) they were
    filled with learning opportunities. I have as yet
    discovered no alternatives and I miss them sorely.

    Another key factor is the capacity of a very few
    obstreporous individuals to flood a newsgroup with
    venemous posts such that it becomes unusable.
    Oh oh! 100baseTx vs the other one (T4?). What were the
    names, Vernon S vs HP man as I recall? I
    think HP lost. Also switched full-duplex token-less
    ring-less token-ring vs switched FD Ethernet. Those
    were the days:)

    Usenet was an experiment and it turns out that
    unmoderated discussion forums have effectively been
    proven not to work - is my view. Any functional
    internet discussion forum needs as far as I can see
    a driven and dedicated individual to manage
    a team of moderators such that the chief
    moderator is able to enforce his vision of the
    purpose of the group. Otherwise chaos
    inevitably seems to reign.

    If anyone has any cisco discussions that they might
    care to recommend I am all ears:)

    Thanks again everyone.
     
    bod43, Oct 16, 2010
    #12
  13. That's the one exception.

    Going from the one device to a device that is the same
    (switch/switch, router/router, host/host). In that case, you use a
    crossover cable.

    Going from one device to a different device is [usually] a
    straight through (router/switch, switch/PC).

    Router to PC is straight through. That's the exception.

    Regards,




    Fred
     
    Fred Atkinson, Oct 16, 2010
    #13
  14. CREAM

    Rob Guest

    This is of course wrong.

    But it has been correctly explained before.
     
    Rob, Oct 16, 2010
    #14
  15. CREAM

    alexd Guest

    Meanwhile, at the comp.dcom.sys.cisco Job Justification Hearings, bod43
    chose the tried and tested strategy of:
    The Netpro forum on Cisco's website weren't all that great when I looked
    about 18 months ago. Plenty of noob questions, not that many experts to
    answer them. Of course, having effective community support probably goes
    against Cisco's commercial objectives ;-)
     
    alexd, Oct 16, 2010
    #15
  16. CREAM

    Rich Seifert Guest

    The "replacement" for Usenet is web-based fora. (Fora is *really* the
    plural of forums.) Of course, the value of such fora is purely a
    function of the knowledge of the participants, and their willingness
    to share. In this regard, they are no different from Usenet.
    It was Fast Ethernet vs. 100VG-AnyLan (IEEE 802.12), with Dan of HP
    (the proponent of VG) and Vernon going at it on a daily basis.
    Fortunately, I was able to maintain a friendship and good working
    relationship with both of them.
    On this we can disagree. Usenet was not an experiment, but a forum for
    interactive communications that thrived for more than 20 years. Things
    change, technology changes, and Usenet is no longer the force it was.
    So be it. But that doesn't mean it wasn't successful. To say that
    would be to say that companies like DEC, that thrived for decades
    before falling by the wayside weren't "successful".
    So a dictatorship is preferable to a democracy, because it is more
    ordered and consistent in its vision and purpose? Chaos can be good.

    --
    Rich Seifert Networks and Communications Consulting
    21885 Bear Creek Way
    (408) 395-5700 Los Gatos, CA 95033
    (408) 228-0803 FAX

    Send replies to: usenet at richseifert dot com
     
    Rich Seifert, Oct 18, 2010
    #16
  17. CREAM

    John Doe Guest

    Anything is possible, but that never happens, at least not for any
    significant length of time. Google Groups (your news server) is
    probably the worst thing UseNet has ever experienced noise wise.
    Works for me.
    Like floods of spam from Google Groups...

    Trying to moderate UseNet would be like pretending you know
    everything about everything there is to know. That is what
    RoadRunner tried to do to its own users before dropping UseNet.

    Yes, some moderated Internet forums are great for commercial
    information specific to a product they sell. And that commercial
    content varies depending on how tied they are to the producer. But
    there is no better place in the world to find anti-BS information
    about anything. And there never will be better noncommercial
    information on any moderated forum.

    UseNet has been made more difficult to get to, but that is not
    necessarily a fault. I suspect that the reduction in use
    corresponds to the worldwide depression as much as it has to do
    with anything else. Also, UseNet is not a friend of the one-way
    media, and someday maybe UseNet will be shut down as an enemy to
    the state. That possibility is the only way I can see for a real
    UseNet demise.

    UseNet is the wisdom and folly of the world. Long live free
    speech.
     
    John Doe, Oct 18, 2010
    #17
  18. Various web fora have been growing for 10+ years.
    Perhaps they lack certain advantages of USENET, but they
    have their own and crucially, have been attracting users.
    Why should either be wrong? The cabel was one of the reasons
    for the decay of USENET. You cannot direct what people wish to
    talk about. You can only silence them, and only in some places.

    Google? USENET has whithered as a result of many factors,
    including an autoaccelerating effect.

    You call that unusable? That was exceptionally mild.

    I have seen moderated fora fail worse. If nothing else,
    the delay stifles them. I am reasonably content to sift
    through 100 irrelevant/uninteresting posts to find one that
    is valuable. I worry that filtering [moderation] technologies
    are certain to generate false-positives, so I sift through 10
    irrelevant/uninteresting posts to find _nothing_ of interest.

    I do not know what you consider chaos. I do not expect harmony
    nor edited journal conformity (thank G-d!) I am quite willing
    to do _research_, including my own filtering.
    I believe the Crisco website has some. They suffer all the usual
    'blog disadvantages with an heaping dose of corporate editing.


    -- Robert
     
    Robert Redelmeier, Oct 18, 2010
    #18
  19. A fair number of questions that previously might have been
    asked here can now be answered through a Google search.

    It seems to me that the reasone for so few posts in
    comp.dcom.lans.ethernet is that there really aren't that many
    questions to ask. Ethernet works! No more questions on the
    3-4-5 rule.

    Some other groups are doing well, others are choked with
    irrelevant posts. I used to read comp.lang.c, but it takes
    too long, and there really isn't much there. People post
    questions about programs presumably written in C (MS Office),
    that are unrelated to C. On the other hand, comp.lang.fortran
    is doing fairly well (and survives with posts that might be
    a little off topic). Comp.dsp also works fairly well.

    I suppose so, but it has to have a real advantage to pull
    people away from something that works. Wikipedia may be
    one that does work, but there aren't so many of them.
    (snip)
    comp.compilers is moderated, and seems to work. The delay does
    make it hard sometimes to have a conversation, but mostly
    it works.

    (snip)
    Try comp.lang.c, at least last time I was there.

    -- glen
     
    glen herrmannsfeldt, Oct 18, 2010
    #19
  20. Good points.
    May I correct that? c.c _appears_ to work. You do not know
    what it could be, what gets dropped or those who get discouraged
    and do not post.

    I bring that up because I have followed comp.lang.asm.x86
    [moderated] and alt.lang.asm [unmoderated] for 15+ years.
    They cover basically the same topic, but have substantially
    different "feel". CLAX86 has had more trouble and gone (mostly
    approve) robomoderation. It is "cleaner" than ALA, but also less
    interesting, particularly with fewer novel ideas.
    I did. ~40 posts/day , average thread depth ~20. Not enough spam
    nor flames :) Seems perfectly reasonable, but as a catchall (wide
    topic), it will necessarily have few threads of interest. So I skim.
    Light work, I've followed NGs with 100s of posts/day and shallow
    threads that are much more work to process. I don't expect my news
    preprocessed or otherwise edited. The editor cannot help but be wrong.


    -- Robert
     
    Robert Redelmeier, Oct 19, 2010
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.