送信者: B Carlson 宛先: Tetsuro Kondo Cc: Alan R. Whitney ; Wayne Cannon ; Dick Ferris ; Dave Graham ; Nori Kawaguchi ; Sergei Pogrebenko ; Misha Popov ; Jon Romney ; Ralph Spencer ; Ed Himwich ; Ari Mujunen ; JUNICHI NAKAJIMA 件名 : Re: VSI-S draft 2 日時 : 2001年4月4日 2:55 Dear VSI Group: I think that it is not constructive to use provocative language during the development of the VSI-S standard. As with anything there is usually more than one way to do it, and our job is to find a solution that meets the requirements, that is robust, and that is as simple as possible. Based on my experience with communications protocols, I suggest, for your consideration, that one way to define VSI-S is by applying the OSI 7-layer (or as many layers as we deem necessary) model to the communications and control problem. This is because, as Nakajima-san has pointed out, TCP/IP already provides a secure communications layer and thus checksums, sequence numbers, and the like are not necessary. On the other hand, I do not agree that a dedicated RS232 link provides an error free connection. We have considerable experience controlling the S2 via RS232, and link errors do show up from time to time (which we've attributed to power glitches produced by cycling heat pumps, antenna motors etc--although the actual source has not be determined). In these cases, the checksums and sequence numbers have proven invaluable. I would strongly oppose any use of RS232 without any error checking or correction layers. So, my suggestion is: 1. The layer 1 protocol is either Ethernet (10 or 100) or RS232 (or some variation of RS232 such as RS422). Exactly which pins for RS232 get used requires a strict definition. It is possible to run a synchronous communications protocol on RS232 (using something like HDLC for layer 2) requiring the use of clock pins, but I presume that for simplicity and generality people want to use an asynchronous protocol with only pins 2 and 3 active. 2. For Ethernet, the layer 2 protocol is IP and the layer 3 protocol is TCP. For RS232, the layer 2 protocol is asynchronous (1 start bit etc. etc) with the packet structure currently proposed (namely: ). The layer 3 protocol defines what is to be done as far as retransmission, packet acknowledgment, timeouts, and retries. 3. Layer 4 and higher protocols (i.e. what the messages actually are and what the DOMs, DIMs etc. actually do with the messages) are the same for both Ethernet and RS232 and are where the bulk of our efforts in defining VSI-S should go. At this point we could diverge from adherence to the OSI 7-layer model and define this as we see fit. If we use the above approach then both Ethernet and RS232 can be used reliably for the VSI without unecessary or unwieldy overhead in the use of either one. Regards, Brent Carlson Tetsuro Kondo wrote: > Dear Whitney-san and Gentlemen, > > Japanese group comments compiled by Nakajima-san regarding VSI-S draft 2 are > asfollows. > > Best regards, > > Tetsuro Kondo > > ------ > > Dear Coordinator, Alan Whitney-san, > > > 5.1 Packet Structure > > Each communication to/from the DTS is an elementary packet of the form: > > > > Japanese group are perplexed with the modified VSI-S Version 2.0. > Especially at the point of the packet structure cited above. > > The VSI-S control sequence is established on reliable > ethernet or RS-232, and already they has own error correction mechanism. > Our intention to establish the VSI-S is pursue a sophisticated > scientific instrument control procedure beyond trifling hardware > difference. > > 1) We would not like to accept any special character before or after > VSI-S original command. > > 2) We can not accept to introduce any checksum with the command > > 3) We agree the merit of the sequence code. But, if one need the > sequence code, it should be a sequential integer as well as unix > process id. > > If one wish to introduce the check sum and invisible headers, > They need explain the reason with occurring error rate in the RS/ether > control communication line. > Our laboratory are originally experts of "communications" and no one > had taken such a way in the reliable internet control. Other High-Energy > Physics group and Remote Sensing groups are interested in the VSI. > It will be disadvantage of the VSI which will take such old-type, > backward command structure. > Our reccommendation is the original simple form. > > [] > > Japanese groups are already introduced real hardware and > software. Especially NAO group is impossible to additional farm-ware > to all existing instruments. We made up our mind to change them if the > new idea to the VSI-S is inovative. But this is not the one. > Regarding the VSI-H, we had introduced the proposed LVDS and TVG patiently. > We will see whether the other groups really need connection between > VLBI systems or just they want to move ahead sole idea. > > Best Regards, > > Junichi Nakajima　nakaji@crl.go.jp > "Communications" Research Laboratory, > CRL,KSRC, Radio Astronomy Applications > 34m Radio Telescope VLBI Group　 > +81-299-84-7145　+81-299-84-7159