Is there a downside to using a subnet mask other than 255.255.255.0?

Question

I just finished a Cisco CCNA video course, so I know just enough about networking to be dangerous. I have a pretty good idea of how IP's and subnets work, but one thing I haven't been able to find out is why, exactly, 255.255.255.0 is the default subnet mask in most of the routers I've seen (admittedly, nearly all consumer-grade and SOHO routers).

Granted, there's not all that much of a reason for a consumer-grade router to use anything else - unless you need more than 254 IP's, or want to do something complex like organizing your network (servers on 192.168.1.X, desktops on 192.168.2.x, etc.)

That got me thinking, are those the only reasons to use 255.255.255.0? What exactly would change for people if the default mask was 255.255.0.0 instead? Is there some downside to using the latter?

Answer 1

Well, sometimes you find some consumer hardware with that mask hard coded here and there. But that's relatively rare and not a deciding factor. Also you want to stick within the predefined private ip ranges, that's the bigger consideration.

Usually common practice for choosing defaults (in my neck of the woods at least) is to use the tightest mask possible and that one is an easy one for humans to deal with even if it isn't the absolute tightest.

You can use whatever you want though just be careful to not open up a range that starts to conflict with internet IPs. There's a list of reserved addresses at https://en.m.wikipedia.org/wiki/Reserved_IP_addresses. E.g. 192.x.x.x/255.0.0.0 is not acceptable for private LAN use.

I know this doesn't answer your question fully. Barring other answers I'll do some research on more concrete reasons why the default is /24 when I'm not on my phone. I strongly suspect it's either tradition, or the safest bet given that unwise users may choose nonprivate ranges.

Answer 2

I guess the answer is easy: Clasful networking is not quite dead (Yes I know, I'm always writing the opposite).

When assigning an IP address many operating systems / tools (Windows, Linux ifconfig) suggest the default mask according to the class of network. As 192.168.x.y is Class-C 255.255.255.0 is suggested.

So if I do something like this on a Linux system:

# ifconfig eth1 10.0.0.1 
# ifconfig eth1:1 172.16.1.1 
# ifconfig eth1:2 192.168.99.1

I get

# ifconfig | grep "inet addr"
inet addr:10.0.0.1  Bcast:10.255.255.255  Mask:255.0.0.0
inet addr:172.16.1.1  Bcast:172.16.255.255  Mask:255.255.0.0
inet addr:192.168.99.1  Bcast:192.168.99.255  Mask:255.255.255.0

As many system behave this way using the default will less likely break things which limits the support costs for the vendors. Also: People are used to this and would complain that the device is to complicated.

Answer 3

Before subnetting there was only Class-based addressing and the smallest networks were /24. So during the transition to subnetting, when the subnet masks were introduced to existing networks, most of them got 255.255.255.0 as being backward compatible. So that was the common thing and is also just a very easy mask to work with, being a round number (in binary) and preserving the Class C semantics of the last octet being the host number and the rest the network number. Using it avoids having to do decimal-to-binary conversion to determine if two addresses are on the same subnet or not.

Another reason I think it persisted besides familiarity and inertia is because there is a lot of broadcast traffic in IPv4 and by definition broadcast traffic goes to every device on the subnet. So in practice, by the time you are done routing and are allocating address to end nodes, you want to keep the number of devices generating broadcasts down to a reasonable number. The Class C subnet 255.255.255.0 is big enough that most organizations won't need more addresses than that and small enough to keep from being overwhelmed by broadcasts.

You do see smaller subnets, like 255.255.255.252, used when handing out publicly routable IPv4 addresses because those addresses are so precious. I did, however, once have an ISP give us 3 addresses with a /30 subnet only to have one of the addresses be on a different subnet than the other 2. That was annoying and probably a mistake on their part, but it was hard to spot because of the math.