Options when facing DHCP Exhaustion?

Question

I've seen twice, an organization's network exhausts its dhcp range.

What to do when a network approaches/hits DHCP exhaustion?

---

The first case I've seen:

Their solution was to put the wifi network on its own subnet with very short dhcp lease times, eg somewhere in the 2hr ~ 8hr ~ 12hr range. This approach means that the daily capacity of the wifi network is that of its dhcp range; but does not handle any more, and can still congest/deny service.

The second case I've seen:

Their solution was again to provision a separate subnet, this time for dedicated devices, but separate from the primary subnet with its dedicated devices as well as occasional wifi usage.

---

Neither of what I've seen in practice seems like a great solution.

Answer 1

If you are running out of IP address in a DHCP pool, you really only have four options to address the problem:

1. Return unused address to the pool faster: this is done by reducing the lease time. This is a popular option for situations like wireless networks at EDUs where they may have hundreds of devices moving in/out of a building as classes change. This will not help if you don't have devices moving on/off the network fairly frequently.
2. Add additional IP addresses to the pool: The simplest way to do this is to check any static DHCP assignments (or reserved ranges of IPs) to make sure they are current and remove any outdated/unused/unneeded assignments releasing those IP addresses back into the pool. If not, or this was already done, two additional ways to do this would be to increase the size of the subnet (i.e. moving from a /24 to a /23) or to add an additional subnet to the pool; both of these options may require addition configuration on the network or individual devices (i.e. ones that may be configured to not use DHCP).
3. Remove devices from the pool: by reducing the number of devices requesting addresses, this should make more available. Remove old/unused devices from the network or if the pool is shared by both wired and wireless devices, ensure that a device isn't requesting an address on both at the same time (allowing one device to take up two IP addresses from the pool).
4. Create a new pool: this can almost be viewed as part of #3 as some devices will need to be moved to this new pool. However it is a bit more involved as it requires additional configuration of the DHCP server (or a second server), typically additional configuration on the router/gateway devices, and usually a new VLAN/subnet which may mean additional configuration changes through the network.

I think that covers just about every solution to this problem that I have worked with in my career. Which actual solution will work best in any given situation is highly dependent on that environment.

Answer 2

If I'm reading what you wrote correctly, both of those solutions appear to be pretty standard. If you're working with a fixed amount of addresses, then you're left with 2 options - NAT or assign only when necessary.

Assuming your routers are capable of handling however many clients you need to NAT, then this option is totally viable. Probably the easiest too. You translate your addresses from an large scope to publicly routable addresses. Changing the DHCP lease duration might even become unnecessary.

If you have devices that absolutely need addressing from a predefined subnet, then you can always hand out this address range to the devices that need it. Oftentimes, phones and printers don't fall into that category. You stand to reduce address consumption by a significant margin if you get stingy with addresses.