Netmasks for dummies 3009

Moe Trin legal stack don't and mask, also

I'm not a detail miner of the RFCs (who is?) but the ones you reference provide the documented experience of inadequate planning that accompanied the huge growth of the internet and the inadequacies of the routing protocols that were meant to deal with it.

While my text searches turned up nothing significant in the way of MUSTs as opposed to "should"s, I think this was a way of accommodating: -- then existing protocols, -- clbuttful addressing and routing, -- the bizarre consequences of subnetting for some networks (which were running lan hosts with public IPs), -- the eventual adoption of CIDR for addressing and routing, -- the now current "standard" required for hierarchical addressing-routing.

(Does that sentence make sense? ;)

Careful reading of some sections also means you have to be really conscious of the differences between network scope and host scope addresses and masks. Even broadcasts (can) have different scopes.

Bottom line is that anyone who does not conform to the current "standard" of contiguous, most significant, 1's bits when defining masks for non-local use, will have very broken routing. What you do on your own (private) network, of course, is up to you ;)

Someone posted a short while back that their ISP gave them a contiguous range of IPs together with a mask that cross a power of 2 bit boundary. Poor guy shares an IP range-mask with others so has to implement a local mask strategy to "correctly" handle a single(?) IP out of the range he was given. Just because his ISP couldn't (wouldn't?) avoid handing out blocks that did not fit29 or28. IIRC, it was a "less than29" block they gave him and it crossed the bit boundary ;-(

Might be interesting to see a working setup not using contiguous 1's bits masks just to see how they deal with the now current "longest match" rule for routing. Wouldn't you like to maintain that network?

regards, prg email above disabled