Stupidity and lies about point to point networks

Stupidity and lies about point to point networks

I've mentioned using /30 assignments for point to point links. It's nothing special, but apparently it was deep magic for whoever set up the links at my school district job. One of the "network engineers" sure had a good time with those things.

Basically, this person took a /24 and carved it up, but not into /30s. Oh no, they were /29s - twice as big. The IPs were allocated like this:

• a.b.c.160 - network
• a.b.c.161 - unused
• a.b.c.162 - unused
• a.b.c.163 - near end
• a.b.c.164 - far end
• a.b.c.165 - unused
• a.b.c.166 - unused
• a.b.c.167 - broadcast

That's a lot of wasted space! I decided to ask what was up one day, and the answer I got was utter garbage. This is what I was told:

It's so the CSU/DSUs can have their own addresses.

Yes, according to this individual, our AT&T/Paradyne boxes were supposed to commit an epic layering violation and reach inside the serial protocol running on those point to point links and pretend to be part of that screwball "network".

Did I mention that these point to point links were running "Bay Standard" and NOT standard PPP? If the odds of a CSU/DSU reaching into a point to point link are already slim, what are the chances they can do that on a proprietary networking scheme? About zero!

So naturally, I asked why it hadn't been used, since more monitoring is a good thing. I only got hand-waving type answers. Even though I came back to it a few times over the years, nothing ever happened.

Years later, perhaps while working on my own "T1 on the cheap" project, I happened to get hold of the manuals for this equipment. I wanted to know just how they would support this kind of mode, and what sort of feats it did to make it work.

Not surprisingly, I found out that the proposed method was utter crap. The way you're supposed to talk to those things is to hook up to the master rack (SLIP and PPP are suggested!), and then it does its own crazy magic between units to spread the network out from there. It reduces the size of the pipe which is available to your router link by grabbing some bandwidth for its own purposes.

Depending on where you look, it's either the "EDL" -- Embedded Data Link, or the "FDL" -- Facility Data Link. All of this is explained at length in the manual, including a bunch of scenarios and even drawings showing how it could be configured. Nowhere in there does it say "blow a bunch of space on ridiculously huge point to point networks".

Of course, considering this is the same network where the IPs were assigned by dividing by 5, it's lucky the thing worked at all.

What a colossal mess.