res@colnet.uucp (Rob Stampfli) (03/12/91)
A few basic questions about uucp over a v.32 modem: 1. Does the addition of MNP1-4 increase or decrease the thruput? Theoretically it would seem that it would increase it by up to 20%, but I wonder if the windowing of "g" protocol would be adversely affected resulting in a net loss. 2. If you are running MNP1-4, is "g" the best protocol to use? If not, what is? -- Rob Stampfli, 614-864-9377, res@kd8wk.uucp (osu-cis!kd8wk!res), kd8wk@n8jyv.oh
dvv@hq.demos.su (Dmitry V. Volodin) (03/15/91)
In <1991Mar12.033518.10492@colnet.uucp> res@colnet.uucp (Rob Stampfli) writes: >A few basic questions about uucp over a v.32 modem: >1. Does the addition of MNP1-4 increase or decrease the thruput? Theoretically > it would seem that it would increase it by up to 20%, but I wonder if the > windowing of "g" protocol would be adversely affected resulting in a net > loss. IMHO MNP1-4 won't help much UUCP-g (if UUCP-g is implemented effectively, of course). Why run one error-correctin proto on top of another using approx the same technique? Aside of your question - MNP5 is worth using. >2. If you are running MNP1-4, is "g" the best protocol to use? If not, what > is? IMHO it is not. You'd better use "f" or even "e", providing, of course, that your modems/drivers treat hardware flow control right. -- Dmitry V. Volodin <dvv@hq.demos.su> | fax: +7 095 233 5016 | Call me Dima ('Dee-...) phone: +7 095 231 2129 |
david@bacchus.esa.oz.au (David Burren [Athos]) (03/16/91)
In <1991Mar15.073701.10005@hq.demos.su> dvv@hq.demos.su (Dmitry V. Volodin) writes: >In <1991Mar12.033518.10492@colnet.uucp> res@colnet.uucp (Rob Stampfli) writes: >>A few basic questions about uucp over a v.32 modem: >>1. Does the addition of MNP1-4 increase or decrease the thruput? Theoretically >> it would seem that it would increase it by up to 20%, but I wonder if the >> windowing of "g" protocol would be adversely affected resulting in a net >> loss. >IMHO MNP1-4 won't help much UUCP-g (if UUCP-g is implemented effectively, >of course). Why run one error-correctin proto on top of another using >approx the same technique? Aside of your question - MNP5 is worth using. MNP classes less than 3 decrease the effective bandwidth of your line. MNP3 gives 108%, while MNP4 gives up to 120% of the native bandwidth. It does this by stripping start/stop bits and running the line synchronously between modems. I've had no trouble running UUCP-g over the top of that. I run UUCP-g connections over 2400bps/MNP4 connections and see actual throughput increases of close to 20%. This is on my home machine that services 15 other nodes. I see similar throughput increases here at work when using the SUN-III (ACSnet) transport protocol. UUCP-g _does_ work with MNP4. However, there are a few variables: The packet sizes used in MNP4 starts at 32 bytes, and if line quality is good this will increase up to 256 bytes/packet. In most MNP modems you can limit the maximum packet size to 64/128/192/256 bytes. I've experimented with various packet sizes, and although I haven't noticed major differences, currently limit it to 128 bytes, as most UUCP-g packets are just over 64 bytes in total size. All the MNP4 connections I use are over local phone lines, not long-distance. Mind you, the line quality between the two is often similar, and local calls are often across multiple exchanges. (This is in Melbourne, Australia) Thus I'm probably not affected by MNP retries very much. Your mileage may vary. IMHO, MNP5 is NOT worth using for UUCP. Most traffic on my UUCP lines are compressed news batches, which conflicts with the compression in MNP5, reducing the bandwidth. It's simple to compress files before sending them (even mail, through compressed-batched-SMTP, although that requires support on the remote host for uncompressing) and that saves on spool disk space. >>2. If you are running MNP1-4, is "g" the best protocol to use? If not, what >> is? >IMHO it is not. You'd better use "f" or even "e", providing, of course, >that your modems/drivers treat hardware flow control right. 'g' _does_ have the advantage of being supported on every UUCP implementation that you may connect to. If you use a protocol that assumes an error-free line you'd better be sure that it _is_ error-free. As Dmitry says, you need to have hardware flow control working. On both ends. UUCP-g works well enough for me over MNP4.... _____________________________________________________________________________ David Burren [Athos] Email: david@bacchus.esa.oz.au Software Development Engineer Phone: +61 3 819 4554 Expert Solutions Australia, Hawthorn, VIC Fax: +61 3 819 5580
tim@dell.co.uk (Tim Wright) (03/18/91)
In <1991Mar15.073701.10005@hq.demos.su> dvv@hq.demos.su (Dmitry V. Volodin) writes: >In <1991Mar12.033518.10492@colnet.uucp> res@colnet.uucp (Rob Stampfli) writes: >>A few basic questions about uucp over a v.32 modem: ... >>2. If you are running MNP1-4, is "g" the best protocol to use? If not, what >> is? >IMHO it is not. You'd better use "f" or even "e", providing, of course, >that your modems/drivers treat hardware flow control right. Do NOT use 'e' on serial lines unless your data is not particularly valuable to you. Just because the modems are error-correcting does not mean that you cannot get occasional corruption at the RS-232 interface between the modem and the machine. This might be rare, but it's important if data integrity means anything to you :-) Tim -- Tim Wright, Dell Computer Corp., Bracknell | Domain: tim@dell.co.uk Berkshire, UK, RG12 1RW. Tel: +44-344-860456 | Uucp: ...!ukc!delluk!tim Smoke me a Kipper, I'll be back for breakfast - Red Dwarf