rja@edison.GE.COM (rja) (05/03/89)
In article <28111@cci632.UUCP>, sz@cci632.UUCP (Stephen Zehl) writes: > - > I'm holding an advertisement from Microcom for their new V.32 modem > that they claim runs at 38,400 bps in MNP Class 9. Actually it only transports data at a raw speed of 9600 bps. The difference is totally accounted for by their compression. Telebit's modems transport at a raw speed of between 14K and 18K depending on telephone line quality. > What's the story here?? The 38.4K only applies to uncompressed data. The Telebit speed of 18K applies to data that has already been compressed, the Microcom speed does not. > Is this the END for TELEBIT TRAILBLAZERS ? Hardly. Most UUCP traffic has already been compressed so Microcom would move it at only 9600 bps, while Telebit moves the same data at more or less 18K. The Telebit is still the better choice for UUCP sites. Moreover, you would need _two_ Microcom modems (one on each end) to use their compression or a V.32 if you were sending already compressed data. The V.32 modems will become more common than at present, but Telebit's T2500 handles V.32 if you need it and besides there are more UUCP sites that already have Telebit modems and so the Microcom wouldn't be able to talk high-speed (9600) to very many other sites (few V.32 sites yet either). I'll be the first to advocate buying a better modem than a Telebit for UUCP/Internet links, but there isn't one out there yet. GE-Fanuc does own several Telebit modems, but these aren't necessarily my employer's views. ______________________________________________________________________________ Internet (vastly preferable) : rja@edison.CHO.GE.COM UUCP (if you've got no choice): ...uunet!virginia!edison!rja ______________________________________________________________________________
jiii@visdc.UUCP (John E Van Deusen III) (05/13/89)
In article <1937@edison.GE.COM> rja@edison.GE.COM (rja) writes: > > Most UUCP traffic has already been compressed so Microcom would move > it at only 9600 bps, while Telebit moves the same data at more or > less 18K. The Telebit is still the better choice for UUCP sites. V.32 modems move data on a good line at 9600 bps in TWO directions at one time. Uucp(1) makes no use of that capability. It is a half- duplex protocol. A TB+ moves data on a good line at a theoretical maximum of 18000 bps in ONE direction. It uses a significant proportion of that bandwidth to implement its PEP protocol, but it is still capable of maintaining a sustained data rate, in one direction, that is greater than is possible with V.32. The TB+ is the modem of choice for uucp(1) communication because the modem implements the g protocol. When the computer in the master role sends out a packet, the modem immediately acknowledges that the packet was received correctly. In this way, subsequent packets can be transmitted immediately. All other modems require that the packet be verified by the computer running in the slave role. If is was received correctly, the acknowledgement must then be transmitted back to the master before a subsequent packet can be transmitted. This requires a lot of time and is required even if the data are transmitted error free by use of the MNP protocol. It is not unreasonable to expect uucp transfers in excess of 800 bytes per second, including all overhead, with a TB+ connected to a 9600 baud interface. This is wall time for transmitting multiple files of precompressed data. I know of no other modem, regardless of raw transmission speed that comes close to accomplishing this. > ... you would need _two_ Microcom modems (one on each end) > to use their compression ... Microcom is a very good and innovative company, and I have used their 2400 baud modems day in and day out with zero problems. That fact not withstanding, there is a very good possibility that MNP compression is destined to become obsolete. For V.42, a form of Lempel-Ziv compression has been specified. This is the reason Telebit could almost immediately claim V.42 compatibility for the T2500; they already support compress(1). -- John E Van Deusen III, PO Box 9283, Boise, ID 83707, (208) 343-1865 uunet!visdc!jiii
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (05/13/89)
>The TB+ is the modem of choice for uucp(1) communication because the >was received correctly. In this way, subsequent packets can be >transmitted immediately. All other modems require that the packet be >verified by the computer running in the slave role. If is was received >correctly, the acknowledgement must then be transmitted back to the >master before a subsequent packet can be transmitted. This requires a >lot of time and is required even if the data are transmitted error free This is incorrect. Uucp sends multiple packets before needing an acknowledgement. If uucp used larger packets, it would work fine (ie, have good performance) on these other modems. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff
ron@ron.rutgers.edu (Ron Natalie) (05/14/89)
> This is incorrect. Uucp sends multiple packets before needing an > acknowledgement. If uucp used larger packets, it would work fine (ie, > have good performance) on these other modems. Or sent more of them before blocking. The problem is not so much the time that it takes the remote host to process the packet, but the delay in the telephone system that makes the larger windows necessary. However, since UUCP doesn't use larger windows, trailblazers are going to win for quite some time.
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (05/14/89)
>> This is incorrect. Uucp sends multiple packets before needing an >> acknowledgement. If uucp used larger packets, it would work fine (ie, >> have good performance) on these other modems. > >Or sent more of them before blocking. The problem is not so much the The problem with just sending more of them is that many of these modems have a large turn around delay that you want to avoid as much as possible. Others have a small reverse channel that the acks for small packets don't fit in. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff