woodman@sumax.UUCP (David Woodman) (11/29/89)
Does anyone know the maximum number of operating systems "hung" off of an X.25 network? 7 is the high water mark so far. P.S. I am posting this for a friend, so if I mangle the terms into utter nonsense, please don't flame me. -- ------------------------------------------------------------------------ David Woodman MAIL: woodman%sumax.uucp@beaver.cs.washington.edu Seattle University #include <disclaimer.std>
panek@hp-and.HP.COM (Jon Panek) (12/01/89)
The actual x.25 specification doesn't address "operating systems" per se. I do believe (from old memory) that the number of bits allocated in the standard message frame for addressee is 8 bits. This would effectively limit the number of addressable sites (on the link level, at any rate) to 256. Jon Panek panek@hp-and.hp.com
howard@cos.com (Howard C. Berkowitz) (12/05/89)
In article <1100@sumax.UUCP>, woodman@sumax.UUCP (David Woodman) writes: > Does anyone know the maximum number of operating systems "hung" off of > an X.25 network? 7 is the high water mark so far. > > P.S. I am posting this for a friend, so if I mangle the terms into > utter nonsense, please don't flame me. I am not completely clear what you mean by "hung," so I will assume you mean what is the maximum number of operating systems containing X.25 implementations which can connect to a single X.25 network. It should be clarified that X.25 is a CCITT Recommendation which has physical, frame, and packet protocols. Clearly, only one machine can plug into each physical layer connector. At the frame layer, the "base" standard allows 127 addresses (or more using an address extension feature), but implementor agreements for the frame layer specify that only two stations will be on each link. Many more users, however, can be multiplexed onto it using packet protocols. At the packet layer, there is a 14-digit address space into which packet layer entities are placed; each address could be for a machine running a different operating system! There are techniques (e.g., upper layer multiplexing, address extension) for getting beyond this limit. There is a 12-bit address space for the number of virtual circuits active on one link. Practical implementations rarely use more than 255 virtual circuits. When I was at TELENET in the early 1980's, I remember generally that there were perhaps 100,000 ports on various hosts of the network; we had around 1000 switches and concentrators inside the public network. These numbers have grown significantly. Even then, there were several private networks using TELENET components which also had thousands of ports (and presumably lots of operating systems). At COS, we routinely do interoperability testing with X.25 networks containing around 8-12 different operating systems. The limit here is primarily the number of computers we can physically put into the lab, and other logistics such as the number of participating companies. I'll be glad to clarify further if you can define your question a bit more clearly. -- howard@cos.com OR {uunet, decuac, sun!sundc, hadron, hqda-ai}!cos!howard (703) 883-2812 [W] (703) 998-5017 [H] DISCLAIMER: Opinions expressed are not necessarily those of the Corporation for Open Systems, its members, or any standards body.
mjhammel@Kepler.dell.com (Michael J. Hammel) (12/05/89)
In article <2050001@hp-and.HP.COM>, panek@hp-and.HP.COM (Jon Panek) writes: > The actual x.25 specification doesn't address "operating systems" per se. > I do believe (from old memory) that the number of bits allocated in the > standard message frame for addressee is 8 bits. This would effectively > limit the number of addressable sites (on the link level, at any rate) > to 256. > I couldn't remember if this was true so I looked it up in a manual I got in an X.25 class I took. Now I'm more confused. According to this text there is a Logical Channel Number of 8 bits and a Logical Channel Group Number of 4 bits. Together these are used to maintain a call. This would then allow upto 4095 maintained calls. To establish a call there are 3 fields: the calling DTE address length field, the called DTE address length field, and the actual address field. By this definition the address length can be any specified length (as long as the address fits between the 4th and 19th octets of the X.25 packet). So in establishing the call the address field is quite large (a bit more than 8 bits), but as far as X.25 is concerned the logical addresses are no more than 8 bits or 12 bits if coupled with the group number. Michael J. Hammel | internet:mjhammel@Kepler.dell.com Dell Computer Corp. | Also: ...!dell!mikeh or 73377.3467@compuserve.com "I know engineers, they looooove to change things" L. McCoy Disclaimer equ standard
daved@tsdiag.ccur.com (Dave Danielson) (12/05/89)
In article <4586@dell.dell.com>, mjhammel@Kepler.dell.com (Michael J. Hammel) writes: > I couldn't remember if this was true so I looked it up in a manual I got > in an X.25 class I took. Now I'm more confused. According to this text > there is a Logical Channel Number of 8 bits and a Logical Channel Group > Number of 4 bits. Together these are used to maintain a call. This would > then allow upto 4095 maintained calls. To establish a call there are 3 > fields: the calling DTE address length field, the called DTE address > length field, and the actual address field. By this definition the > address length can be any specified length (as long as the address fits > between the 4th and 19th octets of the X.25 packet). So in establishing > the call the address field is quite large (a bit more than 8 bits), but > as far as X.25 is concerned the logical addresses are no more than 8 bits > or 12 bits if coupled with the group number. > Mike, when you spoke of 8 bits you might have been thinking of the link level (HDLC) addressing conventions which can be 8 bits in normal mode or 16 bits in extended mode but as far as the x.25 network level addresses go, they are 1-15 digits as prescribed in the CCITT X.121 standard. This is broken down: P NNNN TTTTTTTT PP | | | |- port id (2) | | |-terminal/station number (8) | |-DNIC/Data Network Id Code (4) |-Prefix (1 - sim to the "1" for long distance) Maybe this was unnecessary, but I thought I'd throw it in. -- Dave Danielson - CONCURRENT COMPUTER CORPORATION - Oceanport,NJ 07728 FAX : 201/870-4249 PHONE: 201/870-4137 UUCP : ucbvax!rutgers!petsd!tsdiag!daved or daved@tsdiag.ccur.com