said by jsz0:
A modem has two distinct interfaces: the cable interface and the ethernet interface. The HFC MAC is on the cable/DOCSIS side and used to determine what bits are for your modem and which ones are not. The CPE MAC is the same exact thing except on the Ethernet side.
As far as how an IP gets assigned and the relationship to the CPE MAC; the CMTS relays the CPE MAC to the DHCP server and tells the DHCP server that is requesting an IP. The DHCP server will check to see if you have a static IP set to that CPE MAC, if not it will give you an address out of its DHCP pool but first it looks in its lease database to see if it has already allocated an IP for this CPE MAC -- if so it will send the associated IP out as a DHCP OFFER -- which gets sent out the Ethernet side of your modem, the CPE device gets the offer and assigns itself all the lease information. From there you have functional IP networking from the internet to your CPE.
Alright that makes sense that the CMTS relays the DHCP requests from the cpe to the dhcp server. So I would assume from all this (and doing a little googling on my own), that the typical cable modem boot up goes like so. And please, feel free to correct me if I am mistaken anywhere:
1) Modem powers on, looks for the CMTS, downloads the modem configuration file from the isp's TFTP server.
2) Depending upon the HFC Mac, the modem may end up downloading a particular configuration file (residential customers, business customers, etc.), but there may only be 1 too. It depends on what your ISP has decided. If a modem's HFC mac is allowed to talk on the network, the config file will contain the entry "NetworkAccess 1;" in it. If the value is 0, no further communication can take place (such as DHCP requests) on the ISPs network. However, some ISPs may allow all cable modems access (even non customers)... but simply set an additional setting called "MaxCPE 0;" in a particular config file so that non/blocked customers can talk to the ISPs internal network, but cannot get public IP addresses.
3) If the cable modem has passes authentication via it's HFC MAC (it received a config file with MaxCPE >= 1 and NetworkAccess = 1), then it can pass on DHCP requests for it's CPE devices.
So that's how I think it works, and please correct any mistakes, or feel free to add anything that I may have missed. To sum it all up:
- HFC mac determines whether or not your modem gets access to the isps network. It can also determine which config file your modem will download.
- CPE mac is sent to the dhcp server for a public ip address. CPE mac may also be tied to a static IP if your ISP has it set up like that.
It was hard to find the really technical information about how it works. I couldn't find any official documentation, so much of the information comes from googled websites, forums, etc. I think that is because CableLabs has an NDA with particular vendors on some of the very technical aspects of DOCSIS. I could have been looking in the wrong place too