Device IP Allocation

[Geoff]

Surely having the single scope on DHCP

 

I dispute that. It's benifical to have two scopes if you implement the 80/20 rule to have redundancy with your DHCP servers.

[PiqueABoo]

I said it "raises the bar bit" which is not poor practice compared with leaving the bar on the floor. The approaches are not mutually exclusive, but yes properly configured VLANs, ACLs et al are much, much better. If the equipment supports the latter then I'd seriously consider the pros and cons of implementing that stuff.

 

Meanwhile, IP devices only need a gateway if they need to talk to the outside world and vice versa. Printers, switches, APs and the like do not need to talk to the outside world. And there is no conventional routing in this scenario - IP comms between two 192 devices relies on exactly the same mechanism (ARP) as IP comms between 172 devices.

 

--

 

DHCP redundancy *is* complex. The 80/20 rule relies on the 20 server being reliably slower to respond to requests (it's supposed to be on a remote subnet). The scope split is better based on making an educated guess for "what percentage of your machines are likely to a) have been offline for the lease duration, and b) will get switched on while their local DHCP server is down?".

 

If you've only got one subnet, I'd consider setting up a second service with a few spare addresses but disabling it until needed. You could run both concurrently with a 50/50 split but watch our for bias i.e. significantly more leases being assigned from one server than the other.

[budgester]

(Most) of our printers and older switches (Ones that I didn't install) are currently on an IP range of 192.168.x.xxx ranging from 100 to 300 and a subnet of 255.255.252.0

 

This a valid subnet see reasoning below

 

So normally you would use a standard

 

192.168.*.*

255.255.255.0

 

This would give you 256 networks with 254 hosts (0 for network, and 255 for broadcast)

 

First Network range is 192.168.0.*

Second Network range is 192.168.1.*

Third Network range is 192.168.2.*

Hence 254 hosts per network

 

With a subnet mask of

 

255.255.252.0 you get 64 networks with 1022 hosts.

 

First Network range is 192.168.0.* to 192.168.3.*

Second Network range is 192.168.4.* to 192.168.7.*

Third Network range is 192.168.8.* to 192.168.11.*

Hence 1022 hosts per network

 

Calculations follow:

256 - 4 = 252 Subnet Mask

256/4 = 64 Networks

256 * 4 = 1024 hosts per network

 

Take out your Network and Broadcast address form each network to find out total amount of hosts for each network

 

Subnetting seems like a black art but really isn't once you understand it.

 

So if you wanted to split the 192.168.*.* address range into 16 networks you would

 

256 - 16 = 240 Subnet mask

256/16 = 16 Networks

256*16 = 4096 hosts per network

 

So the reason for using a subnet such as 255.255.252.0 is so that you don't restrict yourself to 254 hosts per network and have to do routing between networks, you would get 1022 hosts per network, which might just enough for a modern school if every student and teacher had a PC connected to your network. Which thankfully they don't have here yet.

 

So how many hosts could you have with a subnet of 255.255.0.0 ?

 

This is commonly used with the private range of 172.25.*.*

 

So 256 * 256 = 65536 hosts per network

 

Note all calculations are done with 256 because in computers 0 is a number to :-) so 0 -> 255 = 256 unique numbers.

 

Am i making sense ?

 

Regards

 

Budgester

[SpuffMonkey]

(Most) of our printers and older switches (Ones that I didn't install) are currently on an IP range of 192.168.x.xxx ranging from 100 to 300 and a subnet of 255.255.252.0

 

300! - surely - you can't have a number bigger than 254?

[sahmeepee]

on a 255.255.252.0 (/22) subnet, you can use the .255 address, because it's not the broadcast address for that subnet, but i'm being pedantic