kenc@nos850.UUCP (Ken Cornetet) (11/15/89)
A question for any NS and DS gurus out there: Given the following network topology: ----------------- ---------------- ---------------- ! ! HDLC ! ! HDLC ! ! ! Node A !________! Node B !________! Node C ! ! 91750 DS ! ! 91750 DS ! ! 91790 NS ! ! ! ! ! ! ! ----------------- --------------- ---------------- | | | 802.3 lan | | | ----------------- ---------------- ---------------- ! ! HDLC ! ! HDLC ! ! ! Node F !________! Node E !________! Node D ! ! 91750 DS ! ! 91750 DS ! ! 91790 NS ! ! ! ! ! ! ! ----------------- --------------- ---------------- Nodes A,B,E,F are running DS under RTE-6. Nodes C,D are running NS under RTE-A. Can node F communicate transparently to node A (ie. can a person on node F do this: CO JUNK::SCRATCH>NODEA JUNK::SCRATCH) ? Ken Cornetet -- I'm not crazy, I just don't | Ken Cornetet work (502) 364-5756, home 937-8826 give a darn! -- D. Duck | 6506 Connie Dr, Louisville, KY 40258
jack@hpindda.HP.COM (Jack Repenning) (11/23/89)
> Can node F communicate transparently to node A (ie. can a person on > node F do this: CO JUNK::SCRATCH>NODEA JUNK::SCRATCH) ? Yes, this works just fine. Getting it configured is a real adventure in mixing incompatible world-views, but it all works. In fact, the single LAN you show could be an arbitrarily large and complex IP catanet, involving lots of LANs and gateways and stuff - say, the whole ARPA-Net. Here's a short discussion on how to set it up: DS Nodes: Configure these guys as normal DS nodes. That is, assign addresses to all your RTE systems from a single, flat address space. (A popular strategy is to call one of them "1", another "2", and so on, but you're free to make up your own numbers:-) The point that needs emphasis here, is that all nodes are numbered from a single space: don't repeat DS nodal addresses on the other side of the LAN, as if they were separate networks. This idea of separate yet communicating networks is an important part of what's new about NS: the DS guys don't grok. Configure the NRV, in the dinit file, using the addresses you've chosen. Mesage Accounting is available among any of your RTE systems, DS or NS. Rerouting is available on any contiguous network involving the links that do rerouting (HDLC, DataLink, X.25), but not across a LAN. Most people want to assign names to their systems, as well as (or rather than) numbers. You assign these names in /system/nodenames, and again, the name space has to span the whole catanet - don't repeat names on the other side of the gateways. NS Nodes: The key to making all this work is the concept of the "guardian node." For each DS node, you have to assign an NS node to act as its "guardian." The guardian's job is to translate NS messages into DS messages (more precisely, whenever NS nodes handle DS services, like file transparency, the whole DS message, including headers, is wrapped in an NS message, and the guardian unwraps it, and sends it on its merry DS way). The chosen guardian must be able to communication with the DS node using normal DS links - HDLC and so forth. For your simple diagram, then, Node C must be the guardian for nodes A and B, while node D must guard for nodes E and F. Now that you've chosen your guardians, you have to explain to the NS nodes which NS nodes are guardians for which DS nodes. This takes place in the NRV entries in the nsinit file (naturally enough). There's an extra column in the NS NRV, compared to the DS NRV, where you list the guardian node. There are several confusing points here: - There are two address spaces going around: the DS one, where everyone takes their numbers from the same pool, and the NS one, where you've drawn three separate networks. With only a few nodes, you would be able to use the same numbers for the DS node number, and the NS node part of a class-C network address. - There are several different name spaces as well: there's the simple, one-word name in /system/nodenames, and there's the three-word nodename used by NS services. Again, you'll commonly assign the /system/nodenames name as the first part of the NS nodename. - Finally, this whole business of a guardian node is not at all obvious. The thing to remember is simply this: your intuition was right, there is a difference between DS and NS, and things wouldn't work automatically. But the NS nodes also speak DS, and that implies that they can translate, as well. But you have to designate a guardian to do the translation for the DS node. I hope that's clear. It's covered in your NS/1000 network manager's manual, and I hope that's clear as well. ------------------------------------------------------------- Jack Repenning - Information Networks Division, Hewlett Packard Company uucp: ... hplabs!hpda!jack or: ... jack@hpda.hp.com HPDesk: Jack REPENNING /HP6600/UX USMail: 43LN; 19420 Homestead Ave; Cupertino, CA 95014 Phone: 408/447-3380 HPTelnet: 1-447-3380 ------------------------------------------------------------- Disclaimer: These opinions are not necessarily those of my employer. In fact, my manager doesn't know I'm . . . Oh shoot! Here he comes! Quick! Pop up that window with code in it! Bye!