That's what I thought, too. But here it is - the official Cisco Press
640-821 Practice Exam (updated from the web site) and it has at least 4 instances of a diagram listing a single router, several switches and 4 hosts. One host has a correct GW address of the router's ethernet interface (10.1.1.254). The other 3 hosts have a GW address of 10.1.1.255 (which I assume is incorrect).
The instructions say to keep taking these practice exams until you get 100% correct - I guess it'll never happen with me.
Unless there's some trick not in the books which allows a broadcast address to be used as a GW address?
remember as well that there is passing the exam, and there is reality.
Theoretically it is possible for the all ones (broadcast) gateway to work. In reality I would home that if you tried to set a device up address 10.1.1.43, mask 255.255.255.0 gateway 10.1.1.255 that it should report an error. Reality says that depends on whoever coded that interface to have thought to check for correct. Should the end system allow it you then have an interesting position...
In this situation, 10.1.1.255 is the broadcast IP address for that subnet and cannot be used by any host.
In any subnet, the highest and lowest IP address are not used. Usable IP addresses = IP addresses in subnet - 1 IP address for the Subnet ID - 1 IP address for the Broadcast IP 254 = 256 - 1 - 1 254 IP addresses remain and your router will use one of them leaving 253 for other hosts.
Example: 192.168.1.0 255.255.255.0 (aka 192.168.1.0 / 24) Lowest IP - 192.168.1.0 (subnet ID) Highest IP - 192.168.1.255 (broadcast IP)
Want to really challenge your brain? Check this out: 192.168.1.0 255.255.254.0 (aka 192.168.2.0 / 23) Lowest IP - 192.168.2.0 (subnet ID) Highest IP - 192.168.2.255 (broadcast IP) In this situation, the IP address 192.168.3.0 is completely valid for a host!!! That is because it is not the highest or lowest IP address in the subnet.
Salutations, Lord Balki CCNA, 1/2 CCNP, 12 years in the field
Want to really challenge your brain? Check this out: 192.168.2.0 255.255.254.0 (aka 192.168.2.0 / 23) Lowest IP - 192.168.2.0 (subnet ID) Highest IP - 192.168.2.255 (broadcast IP) In this situation, the IP address 192.168.3.0 is completely valid for a host!!! That is because it is not the highest or lowest IP address in the subnet.
That I understand but don't know why anyone in the real world would want to subnet (or supernet?) a 192.168 scheme with a /23 mask. I just thought it incredibly sloppy for Cisco Press to issue a diagram with the .255 gateway address when that was clearly reserved for the subnet broadcast value.
Guess it doesn't matter - I did manage to pass the Intro exam yesterday...
It's private addressing, so one is free to do as one wishes.
While I personally prefer to see subnets of /24 and smaller, it can make sens to use a /23.
If you have a building, with logical areas (eg floors or buildings within a campus) that need for example 300 addresses, using a /23 keeps it simple. That way someone on floor 3 has a problem, you *know* immediately which subnet and hence VLAN they are in without having to check.
The best way to do things varies with time. It is not too long ago that the phrase "switch where you can, route where you must" was popular. There used to be a *significant* performance hit if you needed to go via the router from VLAN to VLAN, so you will still see some networks with rather large subnets. I regularly see /20s for users. I know of one large network that has over 2000 users in one VLAN - the "customer" did not understand VLANs and would not let the network bods implement VLANs. If I also add that many were noebook users, and portfast was not enabled you can imagine how bad it was.
Bear in mind that an awful lot of what we consider to be good practice is based upon opinion.