T.R | Title | User | Personal Name | Date | Lines |
---|
8679.1 | what do you expect to accomplish? | NETRIX::"[email protected]" | Farrell Woods | Mon Feb 03 1997 10:43 | 14 |
| Having two adaptors on the same wire/subnet doesn't buy you anything. Pakcets
can be addressed to either adaptor's IP address and of course will be picked
up by the appropriate adaptor and sent up the stack. But IP will use *only*
the first adaptor configured for a given subnet when it sends packets.
The second scenario will work just fine, since you intend to put each adaptor
on its own subnet.
netconfig will let you set up as many adaptors as you like.
-- Farrell
[Posted by WWW Notes gateway]
|
8679.2 | Buys half an adapter | GREGOR::OPP | | Mon Feb 03 1997 13:20 | 11 |
| RE: .1
Well, it sounds like two adapters connected to the same rail
and within the same subnet does buy me � an adapter. A true dual
rail environment is the goal. However, at the moment we're waiting
for the network hardware to assemble the second rail.
Thank you for the technical advice,
Greg
|
8679.3 | Still not clear | NETRIX::"[email protected]" | Dave Cherkus | Mon Feb 03 1997 13:50 | 14 |
| It's still not clear what you want. Are you trying to
increase performance, or availability? If it's performance,
you will probably see a loss, because one controller alone can
saturate the wire, and running the second one will increase
the load on the system because both will generate interrupts
when broadcast messages arrive.
The whole theme of dual-rail comes up time and again. I'm
wondering if the idea comes from some other environment where
dual rail is a win, or if comes from the idea that two has
to be better than one?
Dave
[Posted by WWW Notes gateway]
|
8679.4 | System Qual tests the limits, nes pas? | GREGOR::OPP | | Tue Feb 04 1997 12:51 | 11 |
| RE: .3
In this case, it comes from "If a customer might build such
a configuration, then we (Digital Semiconductor) had better test
it first to see how it behaves." Admittedly, system qualification
and customer's configurations are not always "sensible".
Regards
Greg
|