Hi all,
Here is the situation I find myself in.
Our company has a data center which currently has around 5000 machines operational in it. To keep it simple it's a cryptocurrency mine with dedicated mining hardware, i.e. not general-purpose computers with flexibility.
Right now, we run our network completely using dynamic IP addresses - 172.16.x.x assigned from a central 2960-S switch in our building. From there, each rack of hardware has a WS-C2960S-48FPS-L switch which currently only acts as a switch, if that makes any sense. Pardon my imprecision but I'm rather new at networking at this scale.
So basically we have one "master" switch handing out addresses via DHCP, and that switch is connected to several (60+) switches downstream, which then connect to our hardware.
Our hardware is very basic. It can either get an address assigned to it via DHCP, or it can be manually programmed with a static IP.
Two solutions present themselves:
- Our technicians go down the line and manually program 5000 miners with static IP addresses. No fun.
- Our technicians program each switch to have a static IP assigned to each port on the switch. Which would require only programming the 60 some-odd switches we have on site. Basically the way it would work would be. Port 1 on the switch would be 172.16.1.1, Port 2 on the switch would be 172.16.1.2 and so on - no matter what was hooked up to it. Basically the idea is if we have an issue on say 172.16.1.1 we just go to that physical location, and address it, or switch out the machine for a spare. Basically that port would always be the same address, so we know where it is, regardless of what's plugged into it.
Is the second solution even possible?
I'll sketch out a diagram if it helps.
Thank you!
