| Title: | FDDI - The Next Generation |
| Moderator: | NETCAD::STEFANI |
| Created: | Thu Apr 27 1989 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2259 |
| Total number of notes: | 8590 |
Some interesting information concerning NSC6614 Translation bridges using National Semiconductor Chipset (as described by Nat Semi Application Note 740) One of our customers in London are performing some testing on a ring involving some 7 x NSC6614 (configured 4 ethernets and one FDDI as translation bridges). The ring also includes 2 x NSC6800 DXs (configured with wide area ports - FDDI encapsulation) so that these can "talk" to the native ring, one of the 6614s is set up as a encap/translation gatway. They have been using a Network General FDDI Sniffer and several ethernet sniffers to both generate and observe traffic. They have found that for ring latencey of greater than 5 micro-seconds (or lower with small packet sizes) the 6614s produce duplicate packets if the original was a broad/multicast or unknown single destination, for example, a multicast packet originating on an ethernet segment will be flooded onto the FDDI ring and the 3 other ethernets by the 6614, the packet is then received back on the FDDI ring by this bridge and stripped, however, the packet is now flooded down all 4 ethernets, including the originator. To make matters worse as the packet is seen to have come off the ring the forwarding database is updated to reflect this and so the originator of the multicast is not reachable off its own segment until it transmits again - imagine if this was a boot me request or a DECnet routing update - would the device ever load and how does decnet deal with getting its own update back? Regards, Tony.
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 832.1 | KONING::KONING | Paul Koning, A-13683 | Fri Jan 15 1993 12:13 | 10 | |
Wow. Is that a chipset design error or is the bridge using it wrong? Sounds like it would be useful to look at the chip documentation. Then again... how confident are you of this diagnosis? It sounds like the sort of problem that's too severe ever to make it out of the lab, so I'm a little sceptical. paul | |||||
| 832.2 | Not the first time. | VIVIAN::MILTON | CAUTION - Unresolved Postulates | Mon Jan 18 1993 05:20 | 32 |
Paul, It took a lot of kit to get the ring latencey up to 5 microseconds - 7 x 6614s, 2 x 6800 DXs and a couple of Cabletron concentrators, in fact at one point they were putting this down to a concentrator problem as it only occurred when the last conc. was put in the ring. NSC here in the UK have been trying to emulate the problem but they don't stand a chance because Warburgs (the customer) has all their kit. I understand that NSC Miniapolis (sp?) are starting to do some testing but their first comment was that this looks like a configuration problem - Warburgs are not impressed. As far as chip or implementation goes - I've had a look at Nat Semi's application note and this area seems to be dependant on "additional logic" including the maintainence of a stripping flag - I can fax you a copy of these should you need them. As far as Warburgs are concerned, this isn't getting out of their lab - they've only got the 6614s (which NSC has had to give them free) because NSC could not supply a translation bridge option for the 6800 DX (as promised) due to AMD chipset/implementation problems. Warburgs have demonstrated this problem to me and given the evidence I must agree with their diagnosis. In fact during the lab session we had an instance of a no-owner frame wreaking havoc - this frame was an ethernet broadcast which the transation bridges put on each ethernet once per rotation. The no-owner frame also showed up a mis-understanding about the Cabletron ring puging opperation which turned out to be ring scrubbing - they are now convinced more than ever that they require ring purging and are fast coming to the conclusion they should use Digital's. Regards, Tony. | |||||
| 832.3 | KONING::KONING | Paul Koning, A-13683 | Mon Jan 18 1993 11:43 | 6 | |
I'd be interested in additional information, but I don't really have any significant time to spend on this. So at your leasure... Sure sounds like an excellent sales opportunity for some DECbridge-600s! paul | |||||