subnetting problem

[Enki Doe]

Guys,

i need som help with subnetting.

here is something i dont understand about the subnetting part.

Lets say i got a class B address:

128.32.0.0 /16

I do know that i have 1 network and around 65000 clients.

But when i start subnetting, lets say i got this:

128.32.15.0/30

i can have muti networkaddresses and only 4 hosts. I do know why i can have only 4 hosts, but i am not sure why i can have multiple networkaddresses. Which addresses for the network can i chose between in this specific situation?

many thanks for the help.

[Richard Burts]

I am not clear what you are really asking and whether it is really just a syntax question or whether there is something else in the question. I follow your logic that with 128.32.0.0/16 you have one network with lots of hosts. I am confused when we get to 128.32.15.0/30 and your statement is about having multi network addresses. We both agree that with this mask there are 4 IP addresses in the subnet (and typically we would say that this supports 2 hosts since the typical IP subnet has a network address and a broadcast address). I am not sure what you mean about multi network addresses. Perhaps you mean that with a mask of /30 that there are many network addresses available (because with that mask there are many individual subnets). To be specific about terminology 128.32.15.0/30 is a single small subnet and has a single network address. If you are asking something else please clarify what it is.

[Martin L]

this is called VLSM -variable length subnet mask or CIDR

128.32.15.0/30 is already subnetted and likely gave out to very small businesses by ISP.   But, let's say you are very large corporation and have 128.32.0.0/16 from ISP.  you must divide that by smaller chunks based on hosts in a subnet (sub-network) per several sites.  All sites will get blocks (chunks) based on number of hosts but one of those already subnetted network (block) will be reserved for site-to-site or WAN-to-HQ connections where you need 2 hosts per link. 

Let's use /24 for easy explanation. taking /16 and sub-dividing into /24 gives us  256 networks with 254 hosts per one.  anyway, 1st subnet out of 128.32.0.0/16 will be 128.32.0.0/24, 2nd is 128.32.1.0/24, 3rd 128.32.2.0/24 ......128.32.15.0/24 . .... and so on. 

Now you can take segment 128.32.15.0/24 and sub-divide further down to smaller chunks to get your WAN-to-HQ links: 128.32.15.0/30, 128.32.15.4/30, 128.32.15.8/30, 128.32.15.12/30, and so on.   See, /30 out of /24 gives us 64 links from HQ to remote sites. 

For example,  HQ uses 28.32.0.0/24. Site A uses 28.32.1.0/24, Site B uses 28.32.2.0/24 but none of site will use 28.32.15.0/24 as this one is reserved for WANs links. Those WAN-to-HQ links are as follow:

HQ to Site A link uses 128.32.15.0/30, 

HQ to Site B uses 128.32.15.4/30 ..

HQ to site C uses 128.32.15.8/30  .. and so on. 

if you are small company and got 128.32.15.0/30 from ISP, you must use NAT to go on the Internet;  Also, it is possible that my subnet is 128.32.15.4/30, and we both are part of larger segmentted block assigned for our region by ISP

Regards, ML
**Please Rate All Helpful Responses **

[Enki Doe]

I think i get it. When subnetting, the subnetted host bits can be used as new networks.

Your example:

1st subnet out of 128.32.0.0/16 will be 128.32.0.0/24, 2nd is 128.32.1.0/24, 3rd 128.32.2.0/24 ......128.32.15.0/24 . .... and so on.

With a /16 the third octet will be for hosts by design, but when subnetting /24 the third octet can be completely used for subnets.

Although this confuses me:

HQ to Site A link uses 128.32.15.0/30, 

HQ to Site B uses 128.32.15.4/30 ..

HQ to site C uses 128.32.15.8/30

What will be the ipaddresses i can use in these three examples? /30 bits 30 bits from 32 are for my network. But what are my ip addresses?

[Richard Burts]

Hope I can help resolve your confusion.

For 128.32.15.0/30 The addresses you can use are 128.32.15.1 and 128.32.15.2.

For 128.32.15.4/30 The addresses you can use are 128.32.15.5 and 128.32.15.6.

For 128.32.15.8/30 The addresses you can use are 128.32.15.9 and 128.32.15.10.

If the subnet mask is /30 then 30 bits are network/subnet and 2 bits are host addressing. With 2 bits of addressing there can be 4 addresses. One address is the network address (beginning of the range), one address is the broadcast address (last of the range) and the middle addresses are usable host addresses.

[Joseph W. Doherty]

I hope I don't confuse you, but the concept of "subnetting" (and supernetting) is really part of classful IPv4 addressing. Within classful, blocks of IPv4 addresses, addresses have a specific number of dedicated network bits and host bits. Subnetting is indeed taking some of a classful IPv4's address host bits and using them to extend the number of bits used for the network number. Defining a /24 out of a Class A or B address would be a subnet of the Class A or B address it's part of. e.g. 10.1.1.0/24, but a /24 that's a Class C address is not a subnet, e.g. 192.168.1.0/24.

With classless IPv4 addressing, the Class A, B and C address blocks can be split network/host pretty much any where desired. I.e. in the prior example, both 10.1.1.0/24 or 192.168.1.0/24 are just two different /24s, both have the first number (in the address block) reserved for the network itself, the last number reserved for a network broadcast, and all the other numbers (254 in this case) available for host numbers. One exception is for a /31, which only has two numbers in its address block. They don't support a broadcast address, and the first number serves the role of both a host IP and network number.

BTW, since splitting class IPv4 networks into subnets had been so common, "subnets" and "subnetting" are still used when discussing dividing one (classless) address block into other (smaller) networks, but such networks are really fully independent networks. (With classful routing protocols, there are some "rules" to how they can be used that don't apply to classless routing protocols.)

[Enki Doe]

Thank you for the info. About the /31 part, if there is no broadcast adress or if a network address will be used s an ip address, will that cause any problems?

[Joseph W. Doherty]

Not a problem with routing protocols that support the /31. /31s used for point-to-point links.

[Richard Burts]

Just to expand a bit on @Joseph W. Doherty correct point that /31 is used for point to point connections. The concept of broadcast is meaningful in environments where you can send a single packet out an interface and it can be received by multiple devices. But on a point to point connection when you send a single packet out that interface it can be received by only one device and so the concept of broadcast does not apply to point to point connections.

[Martin L]

lots of videos on you tube about subnetting; please find one that explains well.