Errata

Greetings. Fellow O'Reilly author Jim Van Meggelen here (Asterisk: The Definitive Guide, all 4 versions). You've even mentioned me in your book in chapter 30! (hey thanks!)

I was giving your excellent book a read, and I couldn't help jumping to the T1 chapter (because that's what I do), and I had a bit of a heart palpitation when I got there. The information in there is not as clear as it could be, and unfortunately not entirely correct.

The first and most important thing to understand is that there aren't really any 'types' of T1 circuits. They are all the same, and are only used in North America. A T1 is a physical circuit used within North America to create a DS1 of 24x64kbps timeslots (twenty-four DS0s). It does not define the protocol carried on top of it. Encoding is part of the T1 (AMI is pretty much never used anymore, it's all B8ZS for the last 15 years from what I've seen). Framing is also part of the T1 (and again I've only really ever seen ESF used for the past 15-20 years or so). If the T1 is up, you can start passing data across it. OSI layer 1 and 2, essentially.

Outside of North America, there is no such thing as a T1. The rest of the world uses E1 circuits (very similar to a T1 but instead of 24 timeslots it has 32 timeslots). E1 circuits are also DS1 circuits. Japan has their own circuit type, a kind of hybrid T1/E1 called a J1 (which also is a DS1).

OK, now we get to the protocols you can carry on a T1/E1/J1/DS1 (I always tell my guys, before you start trying to get your protocols going, make sure you've got a T1. Can't count the number of times I've had to help some poor soul who was wasting time troubleshooting a D-channel when he didn't even have a solid T1 going).

But I digress ... back to protocols you might carry across a DS1:

First, PRI: The term 'A PRI' is very confusing to old-school telecom folks, because PRI is a protocol. It's like saying 'A TCP/IP', so it sounds weird to old telecom geeks. PRI is part of the ISDN protocol, which in North America is commonly carried on a single T1 circuit (in fact almost exclusively, but that's not the point). PRI can be delivered in other ways as well (although these days it's pretty rare to see PRI in any format other than 23B+D).

As you noted, PRI stands for Primary Rate Interface. However, the full term should be ISDN-PRI, because the protocol in use on a PRI circuit is ISDN. If you've worked in networking you've heard of ISDN Basic Rate Interface (ISDN-BRI), which is the little sibling to PRI (BRI delivers two B-channels of 64K each and a 16k D-channel ... BRI is actually still used in parts of Europe to deliver POTS service). With PRI, there is no restriction on the number of B-Channels and D-channels you can have. If you run PRI across a single T1 (which of course is extremely common) this could be referred to as a 23B+D PRI circuit. However, there are many more combinations possible. In larger PBXs it was common to have 47B+D, or 46B+2D (so as to have a redundant D-channel), or 92B+2D (four T1s with a primary D and a backup D). Or, believe it or not, when PRI first came out we had a few early systems where we ran 24B-channels across the T1, and ran the D-channel on a 56K serial data circuit! (no I'm not kidding, and yes, this did not work well; good thing somebody got clever and figured out how to send the D-channel across one of the B-channels).

Outside of North America, there is no such thing as a T1, but they still use PRI. They commonly deliver PRI across circuits known as an E1s. Outside of North America PRI is commonly delivered as 30B+D (but can again be all sorts of combinations of nB+nD).

You noted CAS, or Channel Associated Signalling (sometimes named RBS or Robbed-Bit Signalling). There were several different flavours of CAS, intended to behave in a manner similar to older analog PBX trunks (E&M, TIE, DID, etc.). For the most part CAS died out in North America, but in many parts of the world (Latin America for example), they still use those protocols because it is far less expensive. Look up MFC R2 (or R2 signalling), and you'll find yet another use for a DS1.

So, to summarize the nature of this erratum:

A T1 is a physical circuit unique to the US and Canada. The rest of the planet uses E1s (except Japan, which uses a J1). All a T1 does is deliver a stream of bits back and forth in the form of a DS1. E1s and J1s serve the exact same purpose. What you run on the T1/DS1 (E1/DS1, J1/DS1) is up to you.

Popular things to run on DS1s (T1s, E1s and J1s) include the following protocols:
ISDN-PRI (23B+D or 30B+D)
MFC/R2 (very popular in emerging economies if PRI is prohibitively expensive)
CAS FXS (for channel banks)
Clear channel (for whatever you want - in your book this is the type of T1 you have configured - no protocol, just raw data using all channels on the circuit to carry any sort of data)
CAS DID, E&M, etc (pretty much nonexistent for the past 15 years or so, but theoretically still usable)
SS7 (lots of SS7 gets run across DS1 circuits, especially in larger installations where a few hundred DS0s are needed)

Not sure if you're doing another edition, but if so, please feel free to reach out to me and I'll be happy to do the technical review on the relevant sections. Probably not really much needs to be said, since these types of circuits are not gaining in popularity.

Regards,

Jim Van Meggelen

Be careful when deleting access-lists. If youdelete an access-list that is applied to an interface all traffic will be denied.

This is not actually true, All traffic will be allowed when deleting an access before removing off an interface.

However it is still be practice to remove first.

For "dial-peer voice 91 voip" at the bottom of page 566, you apply the SIP-Outbound translation profile. However at the top of the page, when configuring "voice translation-rule 2" you strip leading 9s: "rule 1 /^9/ //".

This works fine for "dial-peer voice 90 voip" because the destination pattern is "9..........". So CME sees a user dial 9, then a 10 digit number and applies the SIP-Outbound translation profile. However for "dial-peer voice 91 voip" CME would see a user dial 911, which matches the destination pattern of 911, then it will apply the SIP-Outbound translation profile and strip the leading 9. So effectively if someone dials just 911, the digits that get sent out would just be 11.

A new translation profile should be created that only includes voice translation-rule 1, to send out 6082222222 as your source/calling number.

Hi, in re-studing routing protocols by this book, and on page 124, regarding rip v2 hops:

it sais the max reachable is 255 and unreachable is 256. Its wrong, On rip v1 and v2 the max reachable is 15 and unrechable is 16.

Regards, awaiting for the confirmation.

On page 124 it says that RIP version 2 has a maximum hop count of 255, but it should be 15.

The error is in the sentance "To add a VLAN to a trunk we would issue the 'switchport trunk allowed vlan 'vlan-number command'.

If this were performed on the switch, ALL VLANs would be REMOVED except the VLAN ID that you just added. This makes for a very bad day..

The correct format should be: "switchport trunk allowed vlan ADD 'vlan-number'.

By using this format the VLAN will be added to the existing group of VLANs already residing on the switchport.

Good read Gary, thanks for taking such a dry subject and interjecting your real life experiences.

Thanks,

Glenn Bradley