subnet question

Discussion in 'Server Networking' started by djc, Sep 29, 2004.

  1. djc

    djc Guest

    I have a general understanding of subnetting. For example given a question
    such as:

    You have this network 192.168.0.0/24. You must create at least 4 subnets
    that can accomodate up to 20 hosts each. What subnet should you assign to
    the hosts on your network?

    This is an actual question I have done on a practice exam. I can break down
    to binary and come up with the 255.255.255.224 subnet mask using simple math
    (2 to the x power - 2 = z, x being bits, z being number of hosts, or
    subnets, depending on what your trying to figure out) But I don't fully
    undertand the application of this? when it says "what subnet should you
    assign to the hosts on your network?" it seems they mean that that one
    subnet 255.255.255.224 should be assigned to ALL hosts on ALL those
    subnets??? is that correct? I was thinking that each individual subnet mask
    would need to be used on each individual subnet??? Meaning if you are taking
    3 extra bits from your hosts bits to use as additional subnets then wouldn't
    each of those subnets have its own unique subnet mask? and wouldn't you need
    to assign each unique subnet mask to the hosts on that unique subnet?? ..
    and NOT have all hosts among the 3 different subnets have the SAME subnet
    mask?

    Hopefully I asked that clearly... Any info is greatly appreciated.
    Thanks.
     
    djc, Sep 29, 2004
    #1
    1. Advertisements

  2. They all would typically have the same subnet mask. the actual subnet they
    are in is "acknoledged" by which "range" their address falls in.
    Simple math? :) How about some simpler math?....The valid subnet host sizes
    are always 128, 64, 32, 16, 8, 4, assuming you are working in the fourth
    octet. So if you want 20 hosts, you subtract 32 from 256 and you get 224,
    whala! there is your subnet mask. If you already have the mask and want to
    know how many hosts are in each subnet, then subtract the mask from 256
    (256-224=32) and you get 32, whala!,...you have the maximum host addresses
    provided by that subnet.
     
    Phillip Windell, Sep 29, 2004
    #2
    1. Advertisements

  3. djc

    djc Guest

    I do like your math! but I need to review subnetting! I just need this last
    piece to completely grasp it.
    thanks for the reply!
     
    djc, Sep 29, 2004
    #3
  4. djc

    djc Guest

    Thanks for the example... its starting to click now. My biggest issue is
    being in and out of this... I just need to follow it through to complete
    understanding and be done with it... actually, and then move on to IPv6...

    thanks again.
     
    djc, Sep 29, 2004
    #4
  5. Unless I misunderstood the question, I think I answered it. Yes, all hosts
    typically would use the same subnet.

    However at the risk of confusing you, there are exceptions, such as
    additional splitting of the last subnet because the last subnet of every
    split is not usable unless it is itself futher split into smaller subnets.
    The best example would be a network using 192.168.1.0 255.255.255.128. This
    splits the network into two subnets of 128 hosts each:

    #1 192.168.1.0 -- 192.168.1.127
    #2 192.168.1.128 -- 192.168.1.255

    However the second subnet (the last subnet) is not usable because it
    contains the "Master Broadcast Address" of 192.168.1.255. Therefore the
    second subnet must be further split into additional subnets which will cause
    there to be a different mask for members of those resulting subnets. So you
    take that second subnet (last subnet) and split it into "64 host" subnets,
    the repeat the process with the "new" last subnet and split it into subnets
    of "32 hosts". You continue to repeat the process untill your needs are
    met or you reach the final "4 host" subnet.

    A "4 host" subnet is used for WAN links, like a T1 with a router at each
    end. In a "4 host" network the first address is reserved as a Network ID
    and the fourth (last) address is reserved as the Broadcast Address which
    leaves you with only two usable addresses. The two addresses are then given
    to the router at each end of the T1.

    The last "4 host" subnet that contains 192.168.1.255, of course, is not
    usable, but at least this time you have only lost 4 addresses instead of 128
    addresses.
     
    Phillip Windell, Sep 29, 2004
    #5
  6. djc

    Dodo Guest

    The practice of losing the last subnet is required for rfc950
    interoperability and only need be followed if non-rfc1812 compliant devices
    are present on the network (highly unlikely).

    http://www.ipprimer.com/addressing.cfm
     
    Dodo, Sep 30, 2004
    #6
  7. djc

    djc Guest

    you did answer my specific question. Thank you. I just meant I feel there is
    one last piece, in 'general' to subnetting that I still need to completely
    grasp the whole subject. I am confused by your additional info though it is
    welcome... I knew that the first and last host addresses were not usable as
    they are the network ID and broadcast addresses, respectively, but I had not
    known that the entire last subnet is not usable.. this, I don't
    understand... yet.

    thanks again!
     
    djc, Sep 30, 2004
    #7
  8. djc

    djc Guest

    thanks for the additional info.. I'll check it out.
     
    djc, Sep 30, 2004
    #8
  9. djc

    Dodo Guest

    Dodo, Sep 30, 2004
    #9
  10. I understand the "classless" concept, but I don't see where that solves the
    loss of the last subnet. With old equipment you even used to lose the first
    subnet, but has been solved by modern equipment (such as the "use subnet
    zero" feature found in Cisco equipment). That article simply said the the
    last subnet isn't lost but never explains why, at least not in any way that
    makes sense. Being the skeptic that I am, I need more solid proof that
    doesn't take 30 minutes to dig through a 100 page of an obscurely written
    document to try to find it.
     
    Phillip Windell, Sep 30, 2004
    #10
  11. djc

    Dodo Guest

    Another source confirming my position:
    http://ralphb.net/IPSubnet/restr.html.

    To what document are you referring when you say, "100 page of an obscurely
    written document."

    I challenge you to find at least one source confirming your claim.
     
    Dodo, Oct 1, 2004
    #11
  12. Was there yesterday. I don't have hours and hours to sit and click "next,
    next, next, next, next" without knowing exactly what and where I am supposed
    to see the articles say something. Being pointed to a specific spot would
    be more usefull.
    This one. http://www.ipprimer.com/addressing.cfm It is too long and
    unclear. It claims that the last subnet isn't lost but never really
    explains why other than to say that it no longer matters if the Network
    portion of the mask is all ones,...which I don't believe since it would then
    be impossible to distingush the Network from the host portion. It also
    makes it sound as if the "lost subnet" issue only occured with a Class C
    address when it actually never really mattered what class the address was.
    The last subnet was always lost regaurdless of the class. If it really means
    the first subnet is no longer lost (like it used to be), then I agree, but I
    don't believe the issue with the last subnet has been overcome. Now if I
    misunderstand the text then that demonstrates my complain that is is
    obscure,..I consider myself a skilled reader, ..at least of English.
    This is the quote:

    "With the removal of the subnet portion, and the liberalization of
    (what is now called) the network prefix, there is no longer a consideration
    of whether or not the bits within the subnet portion are all ones; in other
    words,
    you no longer lose a subnet when you break up what used to be known as a
    class "C" network. "
    I'm not making a claim,...I am questioning one. I'm simply saying I don't
    believe something based on my training and what I have been taught, so it is
    up to someone who believes it to convince me. I am perfectly willing to be
    convinced if a solid case is made.
     
    Phillip Windell, Oct 1, 2004
    #12
  13. djc

    Dodo Guest

    This source says I can break this Class C into six subnets, but this other
    source says you can break the same network into eight subnets. What gives?"
    The short answer is, it depends on which RFC is valid in your environment.

    Perhaps the most significant change on the near side of the 'net under RFC
    1812 is Classless Inter-Domain Routing (CIDR, pronounced "Cider"). Under
    CIDR, the concept of separate "network" and "subnet" portions is now
    considered outdated, and is being replaced by a "classless" addressing
    scheme where addresses can be "subnetted" more freely, without consideration
    of the "class" of address. With the removal of the subnet portion, and the
    liberalization of (what is now called) the network prefix, there is no
    longer a consideration of whether or not the bits within the subnet portion
    are all ones; in other words, you no longer lose a subnet when you break up
    what used to be known as a class "C" network.

    Source: http://www.ipprimer.com/addressing.cfm

    A subnet address cannot be all "0"s or all "1"s. This also implies that a 1
    bit subnet mask is not allowed. This restriction is required because older
    standards enforced this restriction. Recent standards that allow use of
    these subnets have superceded these standards, but many "legacy" devices do
    not support the newer standards. If you are operating in a controlled
    environment, such as a lab, you can safely use these restricted subnets.

    Source: http://ralphb.net/IPSubnet/restr.html

    There it is. I've extracted the gems for you. Two sources supporting my
    position.

    I'm not asking for a technical explanation. I'm just asking you find a
    single internet source agreeing that the last subnet is lost under rfc1812
    subnetting. Is that so difficult?...maybe so.
     
    Dodo, Oct 1, 2004
    #13
  14. The first part was from the same section I quoted, so I knew of it. But even
    with the second section you quoted I do not see where it explicitly states
    that that the loss of the last subnet has been over come. Only the first
    one (that I myself quoted last time) even mentions that a subnet isn't lost
    but does not indicate if it is the first or last subnet and still associates
    it with only a class C network which doesn't make sense. The second quote
    doesn't seem to really deal with it at all. These quotes could easily be
    indicating that the first subnet (subnet zero) is not lost as it once
    was,...which is old news and nobody is questioning that the first subnet has
    been regained and old equipment may not support it. The subnet in question
    here is specifically the last subnet of a group, not the first subnet.
     
    Phillip Windell, Oct 1, 2004
    #14
  15. djc

    Dodo Guest

    A subnet address cannot be all "0"s or all "1"s. This also implies that a 1
    bit subnet mask is not allowed. This restriction is required because older
    standards enforced this restriction. Recent standards that allow use of
    these subnets have superceded these standards, but many "legacy" devices do
    not support the newer standards. If you are operating in a controlled
    environment, such as a lab, you can safely use these restricted subnets.

    Source: http://ralphb.net/IPSubnet/restr.html

    "A subnet address cannot be all "0"s or all "1"s", refers to the first and
    last subnets. "Recent standards that allow use of
    these subnets have superceded these standards..."
     
    Dodo, Oct 1, 2004
    #15
  16. Then, yes, you are asking for a technical explaination at least to one
    degree or another.
    I'm not motivated to bother. You can make it personal if you want to, but
    I'm not. You are the one who dropped in and challenged my original post, I
    just simply haven't accepted it, that's all. And at this point I haven't
    been convinced by your referrences,...maybe later I will, and then maybe
    not,..I'm not that worried about it. I am not opposed to it being the case,
    but I require convincing.

    I went through a year's worth of college preparation for the CCNA exam that
    covered CIDR and is the basis of my posistion. I passed the exam on the
    first shot. I will do it again in a year because it must be renewed every
    three years. I'm sure if these claims are true the newer material by then
    will include it.
     
    Phillip Windell, Oct 1, 2004
    #16
  17. djc

    Dodo Guest

  18. Yes, but I believe this refers to the fact that in the past you could not
    use either the first or the last subnet, but now due to the changes you can
    now use the first subnet (subnet zero) when previously you couldn't. I
    don't see anything specifying that it also allows the last subnet to be
    used. If it is true then fine, no problem, but I want to see something that
    specifically states that the *last* subnet can now be used..... it must
    clearly say the *last* subnet.

    Even in the previously quoted material it said that you no longer "lose a
    subnet" (subnet=singlar=just one). It did not say you no longer "lose two
    subnets". Since originally two were lost, the statement can mean that now
    one is gained back, but one still remains lost.

    I can examine some of the Cisco material I have on CIDR possibly this
    evening, if I have time, and see what is there,...maybe I passed over
    something or misunderstood, ...and then, maybe I didn't. We'll have to see.
     
    Phillip Windell, Oct 1, 2004
    #18
  19. You know, I can't believe no one else has said anything. Surely this is
    something that is well established and has been hammered out long before we
    came along and clashed on it. I'm surprised that some of the more
    experienced guys around here haven't just jumped in and settled it.
     
    Phillip Windell, Oct 1, 2004
    #19
  20. Phillip Windell, Oct 1, 2004
    #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.