Subject: G8BPQ Multiport Packet Switch THE G8BPQ AX25 PACKET SWITCHING SOFTWARE (TheNode) A Progress Report - by John Wiseman, G8BPQ From: CONNECT INTERNATIONAL, January/February 1989 - Copyright 1989 by Radio Society of Great Britain- Reprinted by Permission It is now just a year since I started writing my PC-based AX25 switching software. An outline of the system appeared in CI a few months ago, but for those who didn't see it, I'll summarise its main features: The system was originally planned to be a high performance network node. At the time, the only way of building multiband nodes was to interlink TNC2 (or compatible) TNCs running NET/ROM software via a multidropped async bus running at 9600 bps. This was expensive in both hardware and software, and was limited in both AX25 channel speed, and interport bandwidth, and ithe method of interlinkingg the TNCs (via a diode matrix) made nodes with more than 4 ports very difficult to implement. Things have changed somewhat over the past year (The introductions of TheNet has eliminated the software cost, a variety of TNC2 clones have been produced, and improved interlinking techniques developed), but there is still nothing capable of running very fast links. Also, with the rapid expansion of the network, the need for each port of a multiport node to have its own callsign, and hence entry into the nodes list, has caused the list to get rather large. (The FAT node complex has 8 entries, and the DV system 4). My software allows a multiport node to run with a single callsign, will support AX25 links up to at least 64Kbps (given suitable comms hardware), and eliminate the bottleneck of the async link between ports. The software also allows the user, or more usefully a BBS System, direct access to the Network. This was originally thought to be of less importance than the improved node performance, but the introduction of the multiport BBS systems, and the rather slow introduction of radios and modems capable of high speed operation (not to mention the licencing problems on the band generally regarded as ideal for high speed working (23cms)), has meant the initial installations have been primarily to support multiuser BBS systems. The software will support up to 16 copies of the chief multiuser BBS systems that run in a multitasking environment (WA7MBL and W0RLI), or up to 16 users on the G8UFQ system. (although the BBS systems themselves may not support so many copies - MBL seems to run into trouble with its BTRIEVE files if more than 5 copies are run). This traffic may be trunked over a single radio link (preferably at 9600 BPS on a dedicated link!) to the nearest network node. All BBS ports have the same callsign, which also appears in the nodes list of neighbouring network nodes, allowing the user to connect directy from the local node. CURRENT STATUS The software has been running at a couple of sites since September, and a "Beta Test" stage commenced in mid-December. About six copies are currently running, and are supporting MBL, RLI, and UFQ BBS Systems. The software seems to behave reasonably well, but a few unexplained crashes have occured, so there is still some way to go before it is fit for general release. Also no-one is currently running it as a major switching Node. BENEFITS TO BBS SYSOPS. The system gives two main benefits to BBS operators, and two to BBS users. It allows a multiuser MBL or RLI system to operate with just one radio, instead of needing a separate TNC and transceiver (and band) for each port. Setting up the forwarding system is greatly simplified, as the networking software does away with the need to define each step in the FWD file. The forwarding system should also be more reliable, and the network will automatically reroute round a failed link. The user benefits from being able to call the BBS directly from his local Node, by the Sysop being able to support more simultaneous users, and by not needing to try several different routes and calllsigns to find a free port. FUTURE PLANS Once the current beta test phase is completed, I have a bit of work to do to make the system more like the existing NetRom/TheNet code (eg - Sorting Nodes list into alphabetical order, and implementing the CQ command). I have found a source for a comms card which will run up to at least 256kbps, so I will produce a driver for that, so that I will be ready when very fast microwave links become available. I am also planning a version which can run from PROM, so that a node can be built using a PC motherboard (now available very cheaply) without disk drives. Rather further in the future is investigations into protocols suitable for building a high speed "trunk overlay" network. The existing NetRom system works pretty well at current link speeds and relatively limited total range (nodes in the South East have no knowledge of those in, say, Scotland), but I don't think it is really up to coping with a nationwide network. I think we may end up with regional NetRom-like systems, interlinked by some other system. Any ideas would be welcome! ====================== Mr G. J. Chester G8UFQ ====================== As we were going to press the news was received of the untimely death of Mr. G. J. Chester, G8UFQ, whose BBS software is mentioned several times in this edition of Connect International. Copied by Hank Greeb, N8XX 02-Apr-89 Subject: UK Packet News - Part 2 Final update from Daventry - OR "Who Shot DV" from the series "EASTNODERS" By: John Theodorson G4MTP and Neil Riemer G4JTY - Editors From: CONNECT INTERNATIONAL, January/February 1989 - Copyright 1989 by Radio Society of Great Britain- Reprinted by Permission Well things have gone apace here since we started writing C.I. back in February. When we started the BBS - node links looked like this: AT clone running - +------+------+------+------+ Quadport card (5 serial ports) ! BBS ! BBS ! BBS ! BBS ! EEMS card with 4 Mb memory !Window!Window!Window!Window! 4 TNC''s plus 4 radios ! 1 ! 2 ! 3 ! 4 ! +------+------+------+------+ Software use is WA7MBL ! ! ! ! version 4.31 running under RF Links --> 2m 4m 6m 70cm desQview with four windows ! ! ! ! +------+------+------+------+ Interconnected ! DV2 ! DV4 ! DV6 ! DV7 ! Net/Rom Nodes ! Node ! Node ! Node ! Node ! +------+------+------+------+ Now this was fine and all worked very well but there were may draw backs such as somebody connecting on the 2m port and finding that that port was busin and then in turn trying the various other ports until one was found to be free. Other problems were associated with long convoluted hand- crafted 'forward files' to make the necessary connects to those to whom we forwarded and visa versa. Not to mention the obvious investment in four rigs, four TNC's, Quadport Card and EEMS Card. However as you will see from other articles in this copy of C.I. various software/hardware combinations have become available this year from the brilliant talent available here in the UK. This has revolutionized the concept of running a multiport BBS. The current setup is now: AT clone running - +------+------+------+------+------+------+ 640K Memory ! BBS ! BBS ! BBS ! BBS ! BBS ! BBS ! 1 TNC + 1 Radio !Window!Window!Window!Window!Window!Window! 1.3GHz 9600 Link ! 1 ! 2 ! 3 ! 4 ! 5 ! 6 ! +------+------+------+------+------+------+ Software used G4YFB ! ! running under Desqview ! G8BPQ - TheNode - Software ! with 6 Windows +------+------+------+------+------+------+ ! +-----------+--------------+ ! 1.3 GHz Link @ 9600 Baud ! +-----------+--------------+ ! +------+------+------+------+------+ Interconnected ! DV2 ! DV4 ! DV1 ! DV6 ! DV7 ! Net/Rom Nodes ! Node ! Node ! Node ! Node ! Node ! +------+------+------+------+------+ Now instead of the four non-interlinked windows we have six windows all with the same callsign which will be transparently selected by the BPQ software as each user connects or when the BBS goes into forwarding. The BPQ software also allows direct level 4 connects in the forwarding file. The YFB BBS software runs in a 640K (sic.) window so six windows in a standard 640K machine are possible, and of course now only one comport is used. The many other advantages of Steve's BBS software can be seen in his article elsehwere in this publication. The whole set up runs extremely fast and is very reliable. Gone is the confusion of which radio port is free to connect to, and what is it's callsign. Gone are the days of "BUSY FROM GB7NTS-7:. A simple C GB7NTS from the nearest node will result in a connect to any of the available BBS window. James Miller's (G3RUH) 9600 baud modem performs extremely well and results in the traffic from all six windows flowing much faster than it ever did over the four independant RF links at 1200 baud. As a result of the presentation made by Ian G0CND at the recent Sysops meeting and the subsequent work done by Mike G8TIC and John G8BPQ in coordinting the network we now have a situation where all the traffic flows much more efficiently and reliably around the network. As you see the net result is a far less complicated system running very much more efficiently and effectively. May I recommend to anyone else running a multiport BBS that they give serious consideration to the hardware/software set up described above. Copied by Hank Greeb, N8XX 02-Apr-89