[ih] Fwd: Re: [ipnsig-discuss] Vint's draft
vinton g. cerf
vinton.g.cerf at wcom.com
Fri Sep 6 13:54:13 PDT 2002
per joe's request
>Date: Fri, 06 Sep 2002 11:04:48 -0700
>From: Joe Touch <touch at ISI.EDU>
>Subject: Re: [ipnsig-discuss] Vint's draft
>To: "vinton g. cerf" <vinton.g.cerf at wcom.com>
>X-Accept-Language: en-us, en
>User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.1)
>Could you forward this to the internet-history at postel.org mailing list?
>It addresses exactly the kind of context we're trying to capture there...
>vinton g. cerf wrote:
>>historically, of course, serial digital has been used to convey information with the deep space network (DSN) in the past. There has been an enormous amount of manual labor associated with scheduling and use of the DSN and the IPN, with its Bundle-based protocols, is an attempt to make this process far more automatic. Three way handshakes are tough with such high delays - and a non-starter for flow control. Unique identification of bundles is essential, though, and that's part of the design. Not the same way that SYN did it - we use full date/time strings that would only (!?) repeat if the clocks in the space craft fail. Since that is a non-zero possibility, I suppose we will need to think about how to detect such a problem. The Syn/ack mechanism was designed to deal with 32 bit wrap-around of the sequence numbers for serial octet streams in a TCP connection. There is still some danger on the terrestrial Internet that a wrap around occurs so quickly that it falls within the time
-to-live of the tcp segment. Example, operation of a TCP connection on a terabit channel. Extended numbering in the TCP protocol is intended to deal with that problem. For Bundles, a full date/time, unique source ID and sequence number is used for a similar purpose.
>>Hope this helps!
>>At 06:20 AM 9/6/2002 -0400, Bill Cunningham wrote:
>>>First let me say Vint I'm glad to see in your draft that TCP/IP will be with
>>>us in the future for a long time to come. No ISO OSI here. I read you
>>>mention, this jumped out at me, NAK and ACK of course, but not synchronus
>>>idle. Is there no place in delayed networks of the type you propose for SYN?
>>>Will this network be serial digital transmission like here on earth? Or is
>>>it too slow?
>>>Ipnsig-discuss mailing list
>>>Ipnsig-discuss at isoc.org
>>SVP Architecture & Technology
>>22001 Loudoun County Parkway, F2-4115
>>Ashburn, VA 20147
>>703 886 1690 (v806 1690)
>>703 886 0047 fax
>>Ipnsig-discuss mailing list
>>Ipnsig-discuss at isoc.org
>SVP Architecture & Technology
>22001 Loudoun County Parkway, F2-4115
>Ashburn, VA 20147
>703 886 1690 (v806 1690)
>703 886 0047 fax
More information about the Internet-history