VoIP Guy, You weren't wrong! You said that Q.931 doesn't have sequence numbers, which is true. Q.931 is not LAPD, however
LAPD (Q.921) does have sequence numbers. It looks just like LAPB, LLC2, SDLC, etc. Each side has its own sequencing. They don't have to agree. Each side also tells the other side which frame number it expects next. Check this out: Boston#debug isdn q921 2656.612 TX -> IDREQ ri = 14613 ai = 127 2656.648 RX SABMEp sapi = 0 tei = 64 2656.676 RX RRr sapi = 0 tei = 64 nr = 1 2658.372 TX -> INFOc sapi = 0 tei = 64 ns = 0 nr = 1 It's the NR and NS that you should look at. Each side sequences its frames with the NS number. (I call it the Now Sending number.) Each side also specifies the frame number it expects to receive next from the other side with the NR. (I call this the Next Receive NR number.) A station retransmits if the other side gets behind. There's also a REJ and FRREJ for reporting errors. I don't know for sure that LAPD uses these the same way that LLC2 does, but IEEE 802.2 says this about them: 1 Reject (REJ) -- A station sends a REJ when it receives an unexpected sequence number. 2 Frame Reject (FRMR) -- A station sends an FRMR when it receives an invalid frame or sequence number. There's also flow control with Receiver Ready (RR) and RNR (Receiver not Ready). Howard could tell you more because he knows LAPB in gory detail! ;-) Priscilla At 12:44 PM 11/29/01, VoIP Guy wrote: >I was wrong. > >I looked it up last night and there is a seq. number in the control field of >LAPB, HDLC, and LABD. Both, the sending and receiving stations must keep >the same seq. numbers when transmitting, but I cannot find anything on >retransmission at that layer. But I asked an old IBM guy I used to work >with and he said that SDLC and all the related layer two protocols do >require retrans when bad packets are found or missing. So I would assume >that LAPD layer two is reliable. And as everyone else said, the SS7 >signalling (Q.931) is just control and status messages over D channel. > >And B channel is a different retrans technique, based upon the higher layer >protocols it carries. If an ISDN frame gets corrupt, both channels will be >retransmitted, but by differnt methods. So ISDN D channel is inherently >reliable at layer two and B channel is reliable only if that higher layer >protocol is. > > >""Peter Whittle"" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > I sent this to Priscilla on the topic and she suggested that the group > > might benefit from my response, so here it is. > > > > Priscilla, > > > > I think that you may find it helpful to separate end - to - end data > > transfer from signalling. > > > > Very few L2 protocols offer error correction. The modern approach is to > > require the L1 transmission to provide intrinsically reliable > > communication and hence it is a waste of bandwidth to implement error > > correction both on hop by hop and end to end basis as per X.25. > > Modern WAN digital transmission systems are designed to offer > > transmission error rates of fewer than 1 bit error in 10^9 bits. > > > > On Telco Wan links it is common on this side of the pond to require > > transmission media to offer error rates better than 1 in 10^9 and often > > 1 in 10^11. Indeed the commissioning tests call for fewer than 1 error > > in a 20 minute period on a basic E3 (34 Mb) link and fewer than 1 error > > in 24 hours on International links prior to acceptance from Transmission > > into Networks for operational trunks. That is not to say that links may > > not degrade but if the error rates became worse than 1 in 10^9 it would > > be time for Network operations to call 'holes & poles' (Transmission) to > > fix it. > > > > The fundamental assumptions in both Frame Relay and ATM is that they are > > running over intrinsically reliable transmission media. The low error > > rates being achieved either by correctly engineered transmission paths > > or by the use of significant forward error correction built in to the > > transmission equipment. > > > > ATM, and Frame Relay, implement error correction, or more precisely re- > > transmission in the interface to the signalling protocols. ISDN relies > > on the hop by hop error correction offered by LAPD. However, they tend > > to leave the issue of payload error correction to any high level end-to- > > end protocols being run on top of these L2 Datalinks. > > > > ATM offers no direct protection of payload content, the HEC only > > protects the ATM header. However, some AALs do offer protection if not > > correction of the payload. Even AAL5 - most common for IP has a check > > polynomial (CRC32) to protect the CS PDU. It performs error detection > > but not correction. In the case of Q.2931, SAAL (version of AAL5 to > > carry signalling) will detect faulty PDUs. If you want to look at ATM > > signalling take a look at Q.2931 essentially an enhanced and extended > > version of narrow band ISDN Q.931 signalling. Take a look at the ATM > > forum website. www.atmforum.org > > > > > > Frame Relay has Frame Check Sequence that again will detect faulty > > frames. (Incidentally Carrier Switches tend to drop frames with a faulty > > FCS). Incidentally Frame Relay is sometimes known as LAPF. Take a look > > at the frame relay forum web site. www.frforum.org there are some good > > white papers and the frf's recommendations that you can download. > > > > > > ISDN B channel - is a 64 Kbit clear channel and the network makes no > > assumptions about the contents. It could be any number of data formats > > or indeed it could be 64 K G.711 PCM voice. The most ubiquitous use of > > data over ISDN is to encapsulate it in PPP which is intrinsically multi- > > protocol. However, it is also possible to use HDLC, X.25, Frame Relay, > > or any number of specialist protocols. D channel usage is somewhat > > different. L2 on D channel is Q.921 (as you say also known as LAPD). It > > is perhaps worth pointing out the ISDN signalling is NOT an end to end > > protocol! ISDN signalling only traverses the single hop to the > > signalling processor on the nearest switch. This signalling processor > > then signals to the signalling processor of the next switch and finally > > the signalling processor on the last switch communicates with the far > > end CPE. In Public Carrier Networks the signalling between switches is > > normally SS7 or C7 as it is sometimes known. The D channel is normally > > used for signalling but in the case of Basic Rate may also be used for > > permanently on low speed data services such as X.31 (9k6 X.25 in D > > channel, which uses LAPD for L2 and normal X.25 L3) > > > > Q.931 is used on public networks to communicate with the Carrier's CO > > switch and is fairly primitive in its feature set. QSIG is essentially a > > superset of Q.931 used on private telephony networks to signal between > > PABXs and offers an enhanced set of features such as 'camp on > > extension', 'ring back when free', redirect calls etc. > > > > > > X.25 has hop by hop error detection and correction in L2 - LAPB and also > > end to end in the L3. Sometimes known as 'belt and braces' or 'The Pony > > Express' of data communications. 'We get the data through, eventually, > > no matter how crummy the analogue link is!' > > > > Not being of IBM extraction I am not in a position to comment on SDLC or > > Bisync. > > > > I hope that this helps > > > > Peter > > > > -- > > Peter Whittle ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=27708&t=27568 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
