From: Jay Hennigan (jay@xxxxxxxx)
Date: Fri Oct 27 2000 - 13:38:59 GMT-3
On 27 Oct 2000, Shaun Nicholson wrote in one long line:
> Now this was not on my exam so I anit breaking NDA but when the wording
> says that a DDR circuit (using OSPF) must come up immediately on failure
> of a frame link and drop as soon as the circuit is restored the only
> option in my opinion is to use OSPF demand circuit.
> What do you guys think? would this be acceptable ? I think you cannot
> use the Backup Interface circuit (if your allowed too) but there is a
> lot of delays involved in this. I know you can tweak it but would that
> be acceptable?
You need demand circuit to keep the OSPF hellos from nailing the link up.
But couldn't you also use backup interface to trigger the ISDN "immediately"?
Words like "immediately", "as soon as", and "failure" can cause problems
here.
If the frame link fails with loss of carrier, the line protocol will
drop within a second or two, and within a reasonable approximation of
"immediately" dialing can begin. If the frame link dies at the remote
router, then there will be (default) 30 seconds of lost LMI before the
line protocol drops, and up to 60 seconds before it comes back. This
will be lower than the dead timer to detect a drop regardless of the
OSPF network type.
So, unless there's something I'm missing, the "immediately" part depends
on the nature of the "failure", and dropping the ISDN "as soon as" the
link returns doesn't mean when the cable is reinserted into the router
because of LMI keepalive and full status timers. So, this might be a
case to ask the proctor for clarification of words like "immediately"?
Something like, "By 'immediately', do you mean as soon as the router
detects that the line protocol is down? Are default backup interface
timers adequate to satisfy the words "immediately" and "as soon as"?
I'm assuming that floating statics are not allowed, that would be another
approach.
According to Thomas' OSPF Network Design Solutions p. 208, the OSPF process
is aware of the interface state, and will flood LSA on loss of carrier or
LMI, even before the dead timer expires. So, demand circuit might indeed
be all you need. The LSA caused by the change would be "interesting", and
should bring up the link. Haven't tried this yet.
Would you lose points for both backup interface and demand circuit?
I don't know. It wouldn't hurt anything, the scenario would still work,
and in the real world it probably wouldn't be a bad idea.
> By the way these are all metaphorical questions and I'm again only
> looking for some feedback on this one as I have (and sure other users
> have) found this discussion very useful.
Bring them on! Please! It helps to make us think.
> Again please think about NDA before answering I dont want to upset
> anyone out there. My aim is to make the lab a little bit clearer for
> all involved especially myself.
Agreed, and in my opinion the NDA responsibility is on the person *asking*
the question, so state as you did that it's a hypothetical or a matter of
technique that you're curious about.
If someone by chance happens to ask something that's real close to an
actual lab scenario, a flurry of, "You bum, I'm not going to tell you
because of the NDA" responses not only causes resentment and stifles
honest discussion, but also gives away the fact that the inquiry is close
to home.
-- Jay Hennigan - Network Administration - jay@west.net NetLojix Communications, Inc. NASDAQ: NETX - http://www.netlojix.com/ WestNet: Connecting you to the planet. 805 884-6323
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:25:31 GMT-3