cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2911
Views
0
Helpful
1
Replies

VLSM class B

fiyeha7671
Level 1
Level 1

I'm struggling with my homework as I still don't fully comprehend the concept. I don't necessarily need the full answer but I would appreciate if someone could guide me in the right direction. The scenario is:

The host network is 161. 191 . 200. 0 / 21. We are required to configure a few subnets for hosts with the smallest subnet being 92 hosts and biggest subnet being 870 hosts.

1 Accepted Solution

Accepted Solutions

Joseph W. Doherty
Hall of Fame
Hall of Fame

Network numbering is part of an IP address, it's a group of continuous bits, the left side, or high bits.

A native Class B IP (BTW, classful address is pretty defunct) would be the left half of an IP's 32 bit, or 16 bits, which can hold a numeric value from 0 to 2 to the 16th, less 1, or 65,535 (or 64K).  Such a network has the same range for numbering hosts, although we reserve the first and last number for special purposes.

In classless addressing, a Class IP (like - addressing has other restrictions) network would be a /16.  So, a /21 could be a subnet of a /16.  BTW, subnets and supernets denote networks that are not "native" for classful addressing.  With classless addressing, every network is a network, by itself, but we still refer to subnets if they are within the address range of a larger address block.

For subnetting, we "slide" the demarcation between the network number and the host number to the right, bit by bit.  Each bit move "halfs" the number of possible hosts addresses but doubles the number of networks.

So a Class B (like) network, of 16 bits for both network number and hosts, moved 1 bit to the right makes for two subnets, each a /17, each with 17 bits for the network and 15 bits for the hosts.

Using 10.1.0.0/16 this would create 10.1.0.0/17 and 10.1.128.0/17.  Each of those /17s has a host range of 32K.  Each of those /17s can be, again, subdivided into two.  10.1.0.0/17 becomes 10.1.0.0/18 and 10.1.64.0.18 while the other /17, 10.1.128.0/17, becomes 10.1.128.0/18 and 10.1.192.0/18.  The host range for a /18 is 16K.

Stop and insure you understand what's happening as we "slide" the network/host point along an IP address.

When subdividing address space, we don't have to split every "branch" to the same level.  So, with that initial /16, I might want to retain one of the two /17s, subdivide the other /17 into two /18s or four /19s or one /18 and two /19s or . . .

Again, stop and insure you understand this, as it's needed to solve your problem.

One aspect of the problem is one subnet needs to support 92 hosts.

A /26 has an address space of 64, too small, but a /25 has an address space of 128 (again, though, remember we don't use the first or last address space number for hosts, so a /25 supports 126 hosts).

Another aspect of the problem is one subnet needs to support 870 hosts.

A /24 has an address space of 256, a /23 of 512 and a /22 of 1024 - a winner for supporting 870 hosts.

A /22 is half of your /21, leaving us the other /22 to subdivide, our target being 18 subnets, in total, while possibly only using a /25 (also, a larger subnet could be used too).

/21 has two /22s, one needed for the 870 hosts.

If we subdivide the other /22 into just/all /25s, there would be 8 of them.  Those plus the other /22 would be 9 total subnets, not the 18 required.  Of the eight /25 we cannot subdivide one of them, like the /22 we need it to meet the one requirement, but we can subdivide the other 7.

Doing that would provide 14 /26s, one /15 and one /22, total 16 networks.

Lastly, we can subdivide two of the /26s into 4 /27s.

We then have:

/21 - contains . . .

/22 - for 870 hosts

/25 - for 92 hosts

/26
/26
/26
/26
/26
/26
/26
/26
/26
/26
/26
/26

/27
/27
/27
/27

There's your 18 subnets, one of which supports 870 hosts and another which supports 92 hosts.

If this is still confusing, let us know.  It's the kind of problem, when the "light bulb" finally goes on, it will seem easy.

Another point that can be confusing, although I broke out the subnets into decreasing sizes, you can move around the allocations as long as you maintain the "parent" alignments.

For example, the last four /27s are each under a "parent" /26.  So the parent /26s could be restored and other /26s split.

E.g:

/26
/26

/27
/27

/26
/26
/26
/26

/27
/27

/26
/26
/26
/26
/26
/26

Lastly, in the "real world", you generally don't subdivide address space until you must, also preserving large unused address blocks, for further use.  For example, if you only had to provide the two networks with the explicit number of hosts stated, you might have:

/21 - contains . . .

/22 - for 870 hosts

/23 - for future need

/24 - for future need

/25 - for future need
/25 - for 92 hosts

 

View solution in original post

1 Reply 1

Joseph W. Doherty
Hall of Fame
Hall of Fame

Network numbering is part of an IP address, it's a group of continuous bits, the left side, or high bits.

A native Class B IP (BTW, classful address is pretty defunct) would be the left half of an IP's 32 bit, or 16 bits, which can hold a numeric value from 0 to 2 to the 16th, less 1, or 65,535 (or 64K).  Such a network has the same range for numbering hosts, although we reserve the first and last number for special purposes.

In classless addressing, a Class IP (like - addressing has other restrictions) network would be a /16.  So, a /21 could be a subnet of a /16.  BTW, subnets and supernets denote networks that are not "native" for classful addressing.  With classless addressing, every network is a network, by itself, but we still refer to subnets if they are within the address range of a larger address block.

For subnetting, we "slide" the demarcation between the network number and the host number to the right, bit by bit.  Each bit move "halfs" the number of possible hosts addresses but doubles the number of networks.

So a Class B (like) network, of 16 bits for both network number and hosts, moved 1 bit to the right makes for two subnets, each a /17, each with 17 bits for the network and 15 bits for the hosts.

Using 10.1.0.0/16 this would create 10.1.0.0/17 and 10.1.128.0/17.  Each of those /17s has a host range of 32K.  Each of those /17s can be, again, subdivided into two.  10.1.0.0/17 becomes 10.1.0.0/18 and 10.1.64.0.18 while the other /17, 10.1.128.0/17, becomes 10.1.128.0/18 and 10.1.192.0/18.  The host range for a /18 is 16K.

Stop and insure you understand what's happening as we "slide" the network/host point along an IP address.

When subdividing address space, we don't have to split every "branch" to the same level.  So, with that initial /16, I might want to retain one of the two /17s, subdivide the other /17 into two /18s or four /19s or one /18 and two /19s or . . .

Again, stop and insure you understand this, as it's needed to solve your problem.

One aspect of the problem is one subnet needs to support 92 hosts.

A /26 has an address space of 64, too small, but a /25 has an address space of 128 (again, though, remember we don't use the first or last address space number for hosts, so a /25 supports 126 hosts).

Another aspect of the problem is one subnet needs to support 870 hosts.

A /24 has an address space of 256, a /23 of 512 and a /22 of 1024 - a winner for supporting 870 hosts.

A /22 is half of your /21, leaving us the other /22 to subdivide, our target being 18 subnets, in total, while possibly only using a /25 (also, a larger subnet could be used too).

/21 has two /22s, one needed for the 870 hosts.

If we subdivide the other /22 into just/all /25s, there would be 8 of them.  Those plus the other /22 would be 9 total subnets, not the 18 required.  Of the eight /25 we cannot subdivide one of them, like the /22 we need it to meet the one requirement, but we can subdivide the other 7.

Doing that would provide 14 /26s, one /15 and one /22, total 16 networks.

Lastly, we can subdivide two of the /26s into 4 /27s.

We then have:

/21 - contains . . .

/22 - for 870 hosts

/25 - for 92 hosts

/26
/26
/26
/26
/26
/26
/26
/26
/26
/26
/26
/26

/27
/27
/27
/27

There's your 18 subnets, one of which supports 870 hosts and another which supports 92 hosts.

If this is still confusing, let us know.  It's the kind of problem, when the "light bulb" finally goes on, it will seem easy.

Another point that can be confusing, although I broke out the subnets into decreasing sizes, you can move around the allocations as long as you maintain the "parent" alignments.

For example, the last four /27s are each under a "parent" /26.  So the parent /26s could be restored and other /26s split.

E.g:

/26
/26

/27
/27

/26
/26
/26
/26

/27
/27

/26
/26
/26
/26
/26
/26

Lastly, in the "real world", you generally don't subdivide address space until you must, also preserving large unused address blocks, for further use.  For example, if you only had to provide the two networks with the explicit number of hosts stated, you might have:

/21 - contains . . .

/22 - for 870 hosts

/23 - for future need

/24 - for future need

/25 - for future need
/25 - for 92 hosts