Posted 13 August 2013 - 15:22
So you need to know how many hosts will be required on each segment, and should take in to account growth of that segment.
So for example you wouldn't use a /26 for 58 hosts even though it would work.. Unless your are 100% sure that no more addresses would be needed on that segment in the foreseeable future. Maybe a /25 would be better or even /24 to keep in simple which would allow for growth of that segment even if currently a /26 would work.
Using /16 and /24 segments make it very easy for humans to distinguish the network boundaries x.A.x.x x.B.x.x , or x.x.A.x or x.x.B.x are very easy for a human to see but when you use stuff like /26 it makes it a better harder to spot - say 192.168.14.63 is a broadcast address and that .64 is really a network address (wire) vs a host, etc. That given 2 addresses 192.168.1.62/26 and 192.168.1.65/26 are different segments.
Also you need to understand what space are you limited too, are you working with 10.x.x.x, 192.168.x.x or 172.16-31.x.x or all of the private ranges can be used.. Or are you limited to subset of those, or even a public range you have been assigned.
You might want to consider an admin/infrastructure segment used only for access to your switches/routers/firewalls/wireless controllers/accesspoints even servers might have an admin segment enabled where you can only admin the servers from a specific segment.
Also it possible you might want a storage or backup segment - so that your backup or servers access to storage of your servers don't go over the normal user segments.
Keep in mind that when you create a segment that you will need an IP address for the gateway, broadcast and that 1 address is the actual network or wire, shouldn't use as host address. So for example if I give you 192.168.32.64/26 -- that is the actual network and not a host address on that segment.
Depending on the size of the network you might want to use segments next to each other for a geographic location or building or even part of the building your network is in to allow for route summarization. Say you have multiple /24 networks hanging off a router - you can route to those networks via 1 route entry that covers all of the subnets might be a /23 or /22 or even a /16, etc. But if you hang disjointed segments off the same router you can not summarize that route and would need individual route entries in your table - depending on the size of the network this can lead to very large routing tables that can be a pain to manage and are not as optimal as smaller routing tables.
There are lots and lots of variables that can be taken into account in designing your network addressing.. Don't over use your space just because you feel its almost unlimited.. For example just because you have 10.0.0.0/8 to work with -- doesn't mean you should give every segment 10.?.0.0/16 Your segments should be of proper size to provide for the addresses you need in that segment and allow for growth, but not so large that you might exhaust your space with future growth or would be unable to maintain that scheme if network is expanded via locations or accusation/merger of another network, etc.
What was the description you were given to work with for this network.. Number of locations, types of devices from server/infrastructure, end users, etc. Number of devices for each type, etc.