Determining subnets to choose for VLSM?

sleemie

I'm using the sybex book and in explaing the VLSM to me they don't do a good job of explaining how you determine which subnet ID to use for which networks. They explain about choosing the right block to correspond to the number of hosts you need and they have you draw up a chart and fill in the number of hosts needed, the block to use and the subnet mask, but it doesn't explain how to determine which subnets to use for which networks. To me, it seems like it would make the most sense to start out with all of the smallest blocks first and choose those IDs and then work you way up, but he doesn't do it that way he has you using the smallest blocks in like the 240s and 250s range. there seems to be no ryhme or reason, in the chart he first has you figure the subnet for a 32 block, then a 16, then 16, then 4, 4, 4, 16, 64, 16, 64, 16....wouldn't it make more sense to write up your chart and assign the masks and IDs in ascending order for the following blocks.. 4, 4, 4, 16, 16, 16, 16, 32, 64 and 64 and assign the masks in the order..

lwwarner

Technically Lammle is right, it doesn't matter what order you allocate your blocks in as long as you are careful to start each block at a valid location. (The starting point must always be a multiple of the block size.) However, I don't like his haphazard scheme either. IMO, it is asking for trouble and wastes space.

sleemie wrote:

wouldn't it make more sense to write up your chart and assign the masks and IDs in ascending order for the following blocks.. 4, 4, 4, 16, 16, 16, 16, 32, 64 and 64 and assign the masks in the order..

If you want to be able to allocate your blocks without wasing any address space btween blocks, then you really want to do that the other way around. IOW, allocate the largest blocks first and work your way down in size. This works because the end of a larger block will always be a valid starting point for a smaller block, but not vice versa. For example, in the scheme you proposed, if you start at .0 and allocate 3 /30s you would have to skip a /30 before you could allocate your first /28.

Personally, I usually use low subnet numbers for network infrastructure, and high numbers for user LAN space. So, if I had to do CCNA-style VLSM on a /24 for several PtP links and various sized LAN chunks I would allocate the /30s for my PtPs starting at .0, and then allocate the LAN space, biggest block first, starting at the high end and working my way down.

[In the real world I would actually use /31 masks or unnumbered PtP links whenever possible, but this is CCNA-land.]

All of the above assumes we don't care about summarization. If we do care, then you want to allocate your summary blocks first (again, biggest first) and then do the same VLSM dance inside each summary block.

The bottom line is, you can do this stuff any way you want to, as long as you pick a scheme that makes sense to you and you always start your blocks at a location that is valid for the size of the block.

--Bill