problems with subnet mask

Discussion in 'Network Infrastructure' started by andrewleighton657, Dec 29, 2008.

  1. andrewleighton657

    andrewleighton657 New Member

    7
    0
    1
    hi... i'm new here and i hope someone can help me with this.... you know how it is, it's all going fine in the world of i.t. until you read something somewhere and it kinda blows away everything you thought you knew...

    anyway, i came across something the other day and now am confused with subnetting and subnet mask....

    i quote:

    1 * Class A
    Class A network IP address range = 10.0.0.0 - 10.255.255.255
    For one Class A network:
    Subnet mask = 255.0.0.0
    Network address length = 8 bit
    Computer address length = 24 bit

    16 * Class B
    Class B network IP address range = 172.16.0.0 - 172.16.255.255
    Class B network IP address range = 172.17.0.0 - 172.17.255.255
    Class B network IP address range = 172.18.0.0 - 172.18.255.255
    ...
    Class B network IP address range = 172.31.0.0 - 172.31.255.255
    For each of the 16 Class B networks:
    Subnet mask = 255.255.0.0
    Network address length = 16 bit
    Computer address length = 16 bit

    Alternatively, 16 * Class B combined
    Combined Class B networks IP address range = 172.16.0.0 - 172.31.255.255
    For all 16 Class B networks combined:
    Subnet mask = 255.240.0.0
    Network address length = 12 bit
    Computer address length = 20 bit

    256 * Class C
    Class C network IP address range = 192.168.0.0 - 192.168.0.255
    Class C network IP address range = 192.168.1.0 - 192.168.1.255
    Class C network IP address range = 192.168.2.0 - 192.168.2.255
    ...
    Class C network IP address range = 192.168.255.0 - 192.168.255.255
    For each of the 256 Class C networks:
    Subnet mask = 255.255.255.0
    Network address = 24 bit
    Computer address = 8 bit

    Alternatively, 256 * Class C combined
    Combined Class C networks IP address range = 192.168.0.0 - 192.168.255.255
    For all 256 Class C networks combined:
    Subnet mask = 255.255.0.0
    Network address length = 16 bit
    Computer address length = 16 bit


    okay, this is where it gets a little embarrassing... maybe i should know this but i just cant work it out... why are there two different types of subnet masks for class b and class c (i.e. class b combined 255.240.0.0 and class c combined 255.255.0.0)... what does "combined" mean? i did my 70-291 with cbt nuggets and as far as i can recall it was never taught this way, (they said class b subnet will always be 255.255.then subnet and class c 255.255.255.then subnet).... it was always taught that you cant go into the network address to subnet, which makes sense.... but the info above seems to contradict this - class b 255.240 and class c 255.255.0.0.

    so which is it... and why did i understand it and now i dont..? class a makes sense, 255.0.0.0 but it seems class b and class c use two different versions (class b - 255.255 or 255.240 and class c - 255.255. or 255.255.255)

    something else that puzzles me, if you can use 255.255.then subnet on a class c address doesnt that mean its not strictly a class c subnet but a class b subnet...????

    hope someone can make sense of the above and pass it on....

    thanks
     
  2. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    The problem is the unstated point that the info concerns "Private" addresses *only*. And is written in a very confusing way.
    ( See RFC 1918 )

    The 'combining' is merely another way of saying that two of the private address ranges don't fit into the old classes.

    Note that the private ranges were allocated one from each of the old classes.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  3. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    BTW - welcome to CF!

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  4. andrewleighton657

    andrewleighton657 New Member

    7
    0
    1
    hi harry....

    thanks for the response... i can see what you're saying and i'm glad i'm not the only one who finds it confusing.... i'm just glad you actually understood what i was trying to say as i did waffle abit...

    i think i'll just carry on doing it the way i always have and when subnetting for 172. start the network portion as subnet mask 255.255 and when using 192 start the network address as 255.255.255....

    i could be wrong here and i dont mind someone telling me how i should be doing it, but in all honesty i cant remember when i set up a network with a class b or class c and have always used 10.10.etc, even on home networks....

    thanks for the help
     
  5. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    You will find that most broadband routers sold in the UK (and elsewhere) will use the 192.168 private network.

    IMHO you need to know all three private networks, and also the APIPA network. See RFC 3330 for a fairly full set of reserved addresses, which should normally be blocked on a router.

    And note that 192.168 is not a 'Class C' these days.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  6. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    Also note that on the Internet these days classes are dead and gone. They are legacy things.

    Most ISPs are handing out CIDR type allocations.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  7. andrewleighton657

    andrewleighton657 New Member

    7
    0
    1
    its funny because when i did my server training around 3 years ago i wondered why even have classes to ip's; at the time i worked with mcse's and cisco engineers and no one could answer the question; it does help in the adding process but i doubt very much a router see's an ip and says to itself i have a class b here!

    so it seems that classes have been dropped anyway - that would make sense...

    the problem with ip addressing is the more you think of it the more confusing it gets... it seems i'm the only one who still subnets these days, ethernet used to be limited to 1024 hosts on a subnet but with modern switches i dont think that still stands... most networks i go into only use one router to link the net...

    a loopback address uses anything from 127.0.0.1 to 127.255.255.254... over 16 million ways to ping myself! what a waste of ip's!

    private ip's should have been 10.0.0.0 - 10.255.255.254 and thats it... 172.16 - 172.31 and 192.168.0 - 192.168.255.... what a waste of ip's... they could have been used on the net and freed up the shortage of ip's. the 10. range gives you over 16 million hosts, no one in the world could ever need more that many, use custom subnetting to limit it down.... why even use 172. 192. ???

    i appreciate home routers do use 192 but they dont have to... they could have used something like... 10.10.10.10 with a subnet of 255.255.255.224.... thats still 32 hosts

    anyway, forgive my waffle... just had to get that off my chest!
     
  8. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    You would be surprised. Take almost *any* version of Windows and put in a static IP. Nearly always it will 'fill-in' a suggested mask - and that mask is class-full!

    Classes were specified in the early days because it meant that masks could be 'automatic' and not specified. With the imminent exhaustion of the IPv4 address space it was realized that this was wasteful, so CIDR appeared.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  9. Daniel

    Daniel Byte Poster

    236
    6
    25
    Classes are dead and gone?

    192.168 is not a Class C range anymore?

    Damn you 70-270! You lied to me!

    Aha, I understand ranges, but I'm puzzled by why Classes are dead and gone :biggrin

    Havent come across anything from my 70-270 and my 70-290 book explaining this.

    Thank you in advance :biggrin
     
    Certifications: 70-270, 70-290, 70-291
    WIP: None, but learning SEO/SEM
  10. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    The replacement for the class system is usually known as CIDR. See a quite good introductory article on Wikipedia here.

    The basic pressure was that classes were very wasteful of addresses, and the IPv4 address range was getting closer to being exhausted. The two things that helped extend the life of IPv4 were CIDR and NAT.

    This also meant that it was much easier to allocate small ranges. For example, when I got ADSL in the early days the company gave me a /28 of real IPs to use. It would have been very wasteful to hand out a Class C to each employee.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  11. andrewleighton657

    andrewleighton657 New Member

    7
    0
    1
    yes, i can see your points harry, you've explained classes and why we had them and why we now dont better than anyone else i've asked...

    i think the confusion for this - as what daniel points out - is that 70-291 teaches classes and not at anytime says they're outdated... in fact, the official 70-291 microsoft book implies they're still firmly being used... but like you say, CIDR goes against this....

    a quick question for you... if you had to set up a network - lets say 2000 people - would you still use routers instead of switches and custom subnet...???

    of course you would custom subnet the whole network... lets say a 10. network with 255.255.240 mask (4096 hosts).... but does anyone still divide up the network with routers..? this is something else microsoft preaches but i've never found it in a modern network...

    thanks
     
  12. kevicho

    kevicho Gigabyte Poster

    1,219
    58
    116
    Classes are still useful, take for example a 10.x.x.x network, we know its class A from that ip address, so the subnet mask for a class A network is 255.0.0.0, so if we see the 10.x.x.x address with a subnet mask of 255.255.255.0, we know that 16 bits (and 8 for the address) are being used for subnets, probably not very efficient but an example none the less lol.
     
    Certifications: A+, Net+, MCSA Server 2003, 2008, Windows XP & 7 , ITIL V3 Foundation
    WIP: CCNA Renewal
  13. kevicho

    kevicho Gigabyte Poster

    1,219
    58
    116
    You will still need a gateway device (ie router or pc with 2 nics in and say a rras service) to communicate to different subnets.
    When a PC sends a packet it first checks to see if the destination ip address matches the subnet it is on, any addresses that are not are forwarded to the gateway, if there is no gateway the packet is dropped.

    *edit in your example its basically one big subnet so not needed here, alhough best practice would say keep your subnets as small as possible to avoid excessive broadcast traffic*
     
    Certifications: A+, Net+, MCSA Server 2003, 2008, Windows XP & 7 , ITIL V3 Foundation
    WIP: CCNA Renewal
  14. hbroomhall

    hbroomhall Petabyte Poster Gold Member

    6,624
    117
    224
    It will rather depend on how the network is going to be used. Unless this is a vast call-centre (say) then usually a site that size can be logically broken up into smaller lumps. And M$ protocols start behaving badly when faced with large networks. Which is probably why they suggest dividing them up.

    I wouldn't personally normally allow 2000 people on one network. And if you are going to subnet then that automatically means a router, otherwise there isn't much point.

    Harry.
     
    Certifications: ECDL A+ Network+ i-Net+
    WIP: Server+
  15. andrewleighton657

    andrewleighton657 New Member

    7
    0
    1
    oh.... i wud do the opposite, on a larger network i would divide up and a smaller one i would do as one whole subnet...

    dont forget kevicho you still have a gateway, that is the router between you and the net... i can see what you're saying about a ras server but personally i would put that between the router and the main switch, you still dont have to divide the network

    i guess my question here is ethernet used to be limited to 1024 hosts, this isnt the case anymore, so where routers used to go (internally), is the modern approach now just to use switches...???

    thanks

    p.s. on a large network you would divide up with routers, but i'm not really sure if that makes the network quicker, some would argue switches (not hubs) are now so efficient its quicker to use them instead
     

Share This Page

Loading...
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.