Talk:High-Level Data Link Control/Archive 1

Archive 1

Definition

How can HDLC be defined as a synchronous protocol when later in the page there are asynchronous modes mentioned? Drumroll99 20:19, 9 May 2009 (UTC)

History

So was SDLC derived from HDLC (by removing the balanced modes), HDLC derived from SDLC (by adding the balanced modes), or were both derived from some third protocol? Guy Harris 10:48, 27 December 2005 (UTC)

No, it was the other way around. The original SDLC was asymmetric and when it was taken into the US ANSI committee X3S3 it was extended to a symmetric mode as well.

My question is how can you have a history section with no dates!? When did HDLC work start and when was the first version of the standard approved? —The preceding unsigned comment was added by 24.34.102.194 (talk • contribs) 10:23, 3 June 2006 (UTC).

It is mentioned here that HDLC draft existed in Feb. 1973: https://www.rogerdmoore.ca/PS/CYCLB.html — Preceding unsigned comment added by Litvindev (talk • contribs) 22:46, 10 August 2019 (UTC)

Extra information?

Should the section on bit stuffing be reduced to a simple link to the bit stuffing page?

Yeah, it should IMO. Right now, the information about bit stuffing is there twice in this article, and one description is even wrong. I'll probably get to it in the near future (if nobody disagrees, of course). --Kraymer (talk) 13:18, 18 March 2011 (UTC)

About bidirectional communications

Hello,

After reading the article there is something I still don't understand. How can we get a full duplex communication over a single channel? The protocols that run over HDLC need to speak both ways. Is there some kind of turns based on 7Eh exchanges? —The preceding unsigned comment was added by 85.54.149.35 (talk) 14:28, 1 February 2007 (UTC).

HDLC protocol is designed to run over EIA-422 hardware, which implements full-duplex communication.

The hardware has a dedicated "outgoing" twisted pair of wires and a separate "incoming" twisted pair (4 wires total), typically all bundled in a single cable.

For a particular pair, the HDLC data always flows in the same direction -- it never "turns around".

However, the LocalTalk article implies that the Zilog SCC (a HDLC transciever) both transmits and recieves over the *same* pair of wires. Did that use an extension to HDLC, or some completely different protocol?

--76.209.28.72 19:11, 5 June 2007 (UTC)

RS-232 has TxD, RxD, and ground lines. A transceiver sends data on the TxD line and receives it on the RxD line; signals on both lines are relative to the common ground line. Data can be transmitted in both directions at the same time. I don't know whether that was used for any form of synchronous networking, but it was definitely used for asynchronous communications. Guy Harris 20:08, 5 June 2007 (UTC)

What is "loop-mode" vs. "point-to-point" vs. "multidrop" HDLC?

To answer your questions...

Multidrop HDLC

This is the original SDLC mode, which Normal Response Mode was designed for. There is one master ("primary station") and a large number of slaves ("secondary stations"). the primary's transmitted data is received by all the secondaries, and all of the secondary transmitters are connected together to the primary's receiver, although at most one transmitter is switched on at a time. The primary gives each secondary in turn permission to talk by sending it a frame (commonly an RR S-frame) with P=1. The secondary transmits all the data it has, then ends with a frame with F=1. When the primary sees this, it knows that station has stopped talking, and can poll the next station in turn.

Point-to-point

In this mode, there is only one secondary. You can use NRM, but you cal also permit the secondary to transmit whenever it likes, without waiting for permission from the primary. The simplest adaptation leads to so-called Asynchronous Response Mode, which still preserves a primary/secondary distinction. A more complex version is Asynchronous Balanced Mode, where the two ends are symmetrical.

SDLC loop mode

This is associated with IBM's original SDLC, as it was not part of the HDLC spec. Here, rather than sharing a broadcast but, the primary and the secondaries are daisy-chained, each one transmitting to the next one in line. Normally, stations retransmit the data they receive, delayed by one bit time.

The primary transmits the data it wants, followed by a closing flag and an "end of poll" pattern 0xFE (7 consecutive 1 bits). When a secondary sees this, and has data to send, it converts the 7th 1 into a 0, making it a normal flag character, and sends its data, ending with another EOP. This goes to the next secondary, and so on. finally, it gets back to the primary, which begins a new cycle.

All communication is still to or from the primary, so it does not retransmit the data it receives, which is a copy of the data it transmits followed by the data sent by each of the secondaries in turn.

This permits very rapid arbitration for permission to send, basically one bit time per station on the loop, rather than having to have the primary poll each secondary in turn. There is a provision for devices to join and leave the ring, enabling a bypass relay that can maintain connectivity even when they are switched off.

71.41.210.146 (talk) 00:24, 26 January 2008 (UTC)