4

I was recently graced with a new job opportunity in a large higher ed NE department. This is my first job working with a large network of several thousand users.
Throughout my education, I followed the idea that static IP's are useful from time to time in certain applications and historically used more in smaller networks. To this point, it has been my belief up to this point that DHCP is much more useful in larger settings. For this reason, I was shocked to learn that each of our near 3000 networking devices have painstakingly been allocated static management IP addresses. It was explained that this was easier to handle that DHCP.
I will admit, the scheme for the addressing is nifty. The second octet is the device-class/network identifier, the third octet is the building identifier (we have some odd 200 buildings on campus), and the fourth is the device identifier.
So the scheme is understandable, but still seems like a lot of numbers to remember; imagine I am looking for a Xirrus array in say the Cafeteria on St. Phillip's Street. Yes DNS could solve this, but the IP's do sometimes get changed and that could result in a nightmare if one is changed and not the other.

The solution to me seems a DHCP/DNS combo.

So to sum up the question, from a management standpoint, why is it a good idea to use Static or DHCP addressing on equipment?

  • How does the company "document" what it is doing? is there a database of every thing they use/do or is it just what you write down? I have seen how large and small companys often ignore this and cause gread ammount's of extra work later on as no one knows what to do. If the company has some kind of helpdesk software etc that has immaculent records of how to install every app used and all the hardware and ip information then the results "may" be different. – Kendrick May 25 '14 at 3:37
  • Every bit of information is documented. We actually have a database that keeps records of when documents were edited. – lampwins May 26 '14 at 22:59
2

I’ll keep this post about your situation specifically.

First and foremost, managing over 3,000 devices with static IP addresses is just plain foolish. DHCP is a well-known and established protocol that solves the tedious task of manually assigning and fixing manual IP addresses. If they are that concerned about making sure they have control over every IP address on their network, set them as DHCP reservations. Even from a security standpoint, you don’t stand to benefit that much from eliminating DHCP servers.

By sticking with this route, you will lose:

  • Change management capabilities
  • Customer insight

The big hitter on this list is change management capabilities. If you need to swap IPs around, you really don’t have a good way to do this; and that makes (or will make) your life arduous. If your teammates aren't on board with it and want to stick with their old ways, try to rope in your management and inform them of what a colossal waste of time it is.

Regarding your nifty IP scheme; you will still be able to maintain that; If anything, to a greater degree. IPAM gives you the capability to drill down into smaller and smaller subnet ranges, making it seemingly easy to implement in your case. This will allow you to cascade further and further down into the building/device class you desire.

  • foolish, not so much. I've managed networks with tens of thousands of address assignments. (Hint: ISPs do this every day, and have for decades.) – Ricky Beam May 24 '14 at 15:55
  • That's different, because that's out of necessity and is, by virtue of being an ISP, a relatively static setup. In a customer focused environment, where devices move frequently, that doesn't make much sense. When I see quotes like "Cafeteria on St. Phillip's Street", I imagine they don't put an IP out there and forget about it for years. – Ryan Foley May 24 '14 at 16:04
  • It's still a "managed asset". And ISPs can, and do, use DHCP assignment of endpoints. (eg. old bellsouth dsl was dynamic link address, if you had a static block, they were aimed at that dynamic when you logged in.) – Ricky Beam May 24 '14 at 16:19
3

I suppose it boils down to what you mean by networking devices. Routers, switches, firewalls, access points, etc., etc. do need to have fixed, known locations. It's a bad idea to rely on DHCP and DNS -- even where "sticky" addresses are used.

One network fault is all it takes to learn how bad an idea that is... say your network is broken somewhere. You (in fact, the entire engineering department) happen to be on the side that cannot reach DNS, and probably DHCP too. If all of your gear has random, unknown DHCP assigned addresses, you'll have no map, and no access to create one. If your workstation is portable, off, etc. you may not have an address from which to work. Various networking devices may start dropping off the network as their leases expire, other devices (like your laptop) coming online for the day won't have an address, rebooted equipment won't have addresses... A failure of the DHCP server will net a similar effect. A failure of the DNS server(s) will leave a mess of old, invalid registrations, and a bunch of holes where new registrations didn't happen -- nothing I know of will retry.

1

The answer to your question is mostly opinion, which is frowned upon on this forum. But I'm willing to bet real money that the majority of people here will agree with me.

I suppose that for some people, spending lots of time with Excel spreadsheets, keeping track of multiple spreadsheet versions, and the intellectual challenge of troubleshooting duplicate address problems are all enjoyable, but that's really no way to manage a network. I won't speculate on why DHCP was never implemented (cough fear *cough), but DHCP will help you manage your network more effectively by freeing you up from doing a lot of mundane tasks that can and should be automated.

I should point out that not everything will use DHCP. Your routers, switches, firewalls, some servers, will need static addresses. But the majority of devices, especially end user PCs, wireless devices, printers, etc. should use DHCP.

What you really need is an IP address management application, or IPAM. There are commercial products as well as freeware that manage both DHCP scopes and static addresses.

You don't have to give up your nifty addressing scheme either. You can create DHCP "scopes" for each of your locations and device types.

Certain devices that need stable addresses (like printers) can use DHCP reservations, so you can manage their addresses from a single application. That saves a lot of "running around" time.

In addition, you can use dynamic DNS, that syncs DHCP and DNS so that when an address is leased, a DNS entry is automatically created. Your Xirrus AP will always have the same name no matter where it is.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.