2

I'm managing a /27 network. Obviously I cannot assign the network address to any host, but I found assigning the broadcast IP to a host seems to be OK. The host with the broadcast IP of the network as its host IP address seems to be without any problem to access the network. Now I'm not sure whether doing this could cause any hidden problems? Thanks.

2 Answers 2

6

It can introduce performance problems since packets to the broadcast IP address are broadcast to all hosts on the segment. This additional traffic which is intended for only one host but send to every host can negatively impact the performance of those hosts needlessly receiving that traffic.

Some switches can rate limit broadcast traffic. If your switch does that, the host to which you assigned the broadcast IP address may never be able to receive data from the network at full speed.

Moreover some of the other hosts which receive packets destined for the broadcast IP might send error responses (ICMP errors or TCP RST packets) when receiving unexpected traffic. If any host does so, it can cause broken connectivity for the one host attempting to use the broadcast IP.

If being able to use this one additional IP address from your assigned range is critical to you, there is a few other steps you could take.

First of all start deploying IPv6 immediately. It is the only long term solution to IP address shortage.

Secondly if you want to squeeze every IP address out of a prefix which has been routed to you, you can do it at a cost of making your network configuration much more complicated.

Broadcast address and network addresses are not assigned the usual way if the netmask is 255.255.255.255. Additionally the gateway and broadcast address does not have to be from the same prefix as the hosts.

By taking advantage of that you can reconfigure your router and every host on the network. That will give you a 10% increase in the number of usable addresses on your segment at the cost of a 200% increase in complexity of the configuration.

Overall such a reconfiguration doesn't seem worth it.

5
  • Thanks @kasperd. So does it mean I should leave this address not used at all?
    – Qian Chen
    Jul 19, 2015 at 10:51
  • @ElgsQianChen It is possible, but I would not recommend it. Instead I would recommend that you move ahead with IPv6 deployment.
    – kasperd
    Jul 19, 2015 at 11:23
  • Thanks. I am still trying to educate my self with IPv6 knowledge. But I found it quite difficult. My brain seems not to be ready for the IPv6 yet.
    – Qian Chen
    Jul 19, 2015 at 12:45
  • 1
    @ElgsQianChen If you have seen an IPv4 network without any address-shortage-workarounds, then an IPv6 network will look very familiar. What gets many people confused at first is textual notation for IPv6 addresses and in particular the rules about abbreviation. Don't worry too much about that. Understanding IPv6 is not about understanding that notation but about understanding of how the network communication works (where the textual representation isn't used anyway, since addresses are sent in binary as 128 bits).
    – kasperd
    Jul 19, 2015 at 13:01
  • 1
    @ElgsQianChen And don't try to understand transition mechanisms like all sorts of tunnels and NAT before you understand IPv6 itself. That means start with native IPv6 first because that is easier to understand than tunnels.
    – kasperd
    Jul 19, 2015 at 13:02
1

While I don't recommend this, the above answer is not quite correct. Ethernet does not broadcast based on IP, but rather MAC address, and only frames with FF:FF:FF:FF:FF:FF as the destination will get broadcasted across the network, regardless of the IP.

In theory using the broadcast IP should work without broadcasting all its traffic. For instance 192.168.0.1 wants to talk to 192.168.0.255. It will ARP and get the MAC address for .255. It will then create a frame with the .255 MAC address. When it puts it on the network, the frame will be switched to only the port where .255 lives.

1
  • But the problem is, that other hosts think .255 is the broadcast address and might therefore create frames with FF:FF:FF:FF:FF:FF instead of using ARP.
    – JojOatXGME
    Oct 4, 2020 at 12:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .