1

There is a thread from about 9 years ago that I was reading when looking into whether we should change the servers in our environment to DHCP.

Here is the Link.

DHCP addressing vs Static addressing for Servers

It seems that most of the common concerns can be mitigated quite easily these days.

-Have a failover DHCP cluster for redundancy -Use DHCP Reservations -port and network security to prevent rogue dhcp servers

It seems that there is a real divide between the static people and the DHCP camp, and I would like to get a better understanding of both sides.

What are some of the reasons you use DHCP for servers? or What are the reasons you have chosen not to?

I feel that for our environment 500+ servers and growing, that the centralized network management and and the availability of using DHCP with reservations would be a huge benefit.

2 Answers2

2

I feel that the centralized network management is a huge plus, but there are people here who are dead set against it.

I think a lot of people that want/need central management have moved towards the configuration management (Ansible/Chef/Puppet/DSC/etc) way of doing things. So an individual computer may be configured with a static address. But that static address was assigned was done by some configuration management tool.

I also strongly question your premise that DHCP, or some other dynamic way of allocation of addresses is not being used. I think most of the hundreds of cloud-based services and local clouds operate of some dynamic method of address assignment. Perhaps not DHCP, but some dynamic allocation.

Zoredache
  • 133,737
1

Why aren't more people moving to DHCP for their server environment? Is it because that's the way it has always been done, and people don't like change?

Obviously we can't know why people don't move. I'll reply here because I had this "conflict" many times with many customers in my professional career.

I'm one from "DHCP camp". Where the customer has not enough skills to argue, I'm always been fine with DHCP. Over 20 years in my experience I've met many CTOs arguing with this practice.

Most of the times they used to sustain that it's still one more dependency. Despite any redundancy, despite the fact that without DHCP they'd have the servers up and no clients able to use them, they wanted all servers to be static.

Each time they need to push a new route someone from their ICT still calls me: "Server XYZ doesn't connects to the new subnet. Is it a firewall matter?"

I ended up thinking that most of the managers I've met just can see "people will work one minute less" and not "how much it costs to handle my system".

Not much to argue here on my side, smile, work, thank. Customer's always right.

Marco
  • 1,864