wlm@archet.UUCP (William L. Moran Jr.) (03/12/89)
In article <53244@yale-celray.yale.UUCP> Owens-Christopher@cs.yale.edu (Christopher Owens) writes: >Would someone be kind to answer a few naive questions about modem >technology. I'm interested in communicating between a Macintosh-based >terminal emulator and a couple of unix sites, for mixed interactive >and file transfer use. > >1) What do the various levels of MNP mean? OK, (hi Chris). Here goes: MNP 2: Uses async framing; throughput < baud rate MNP 3: Uses sync framing removes start and stop bits; gets something like 25% better perf than MNP 2; throughput marginally better than baud rate MNP 4: Uses sync framing plus something called "Adaptive Packet Assembly"; and "Optimized Data Phase"; I think this means it varies packet size based on external factors. throughput about 10% better than MNP 3 (~2800 bps with a 2400 bps modem) MNP 5: Uses data compression. This does about 1.8x as well (when with MNP 4 as MNP 4 does alone). MNP 6: Auto reliable; allows modems to negotiate as to the best speed they can both manage. Also statistical duplexing. 1/2 duplex. MNP 7: Better compression than MNP 5 (I think when used with MNP 4 it does about 2.2x as well as MNP 4 alone). MNP 9: MNP 7 Plus V.32 (i.e. full duplex). As far as I know, MNP is owned by microcomm and only their modems use MNP higher than 5. > >2) If I am talking to a Unix site over modems that offer some kind of >automatic speed selection, I assume I want to tell the Unix host and >my terminal that we are talking at -- say -- 9600 baud and let the >modems do their thing. Is this correct? Can MNP modems support one >speed between the terminal and the modem, and anothere speed over the >long-haul (telco) lines? > Yes, above a certain mnp level (I'm not sure which), you set a speed to the modem (for example tip at 19.2k or have a getty running at 19.2K), and the modem then transmits to the serial port at this speed while deciding what speed it will send to the other modem with; this can be hazardous. For example, the V.32 MNP 9 modem I have will connect to another and get throughput better than 9600 (19.2 approx) and if you are connected to the modem at 9600, it's buffer overflows eventually. Actually, I suspect that this is not a feature associated with MNP levels, but I'm not sure. >3) I assume that these modems have a variety of parameters that need >to be twiddled. Is this done via some superset of a Hayes-like >command sequence, or is there some other method that must be used to >tell the modem what to do. > Yes, some of these suckers have about 200 pages of commands and explanations :). Seriously though, they use extended Hayes commands (mostly) in addition to proprietary command sets (ugh). >4) The (just-released) Practical Peripherals 2400 SA/MNP has been >recommended to me. Anybody have any opinions on it? No, sorry. Bill Moran -- moran@cs.yale.edu wlm@ibm.com uunet!bywater!acheron!archet!wlm Idealism is fine, but as it approaches reality the cost becomes prohibitive. wfb