[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: EOS WG Minutes - Minneapolis
To pick on one particular topic:
> From: Lauren Heintz[SMTP:email@example.com]
> Sent: Thursday, March 29, 2001 6:19 PM
> To: Juergen Schoenwaelder
> Cc: firstname.lastname@example.org; email@example.com
> Subject: Re: EOS WG Minutes - Minneapolis
.... snip ....
> Personally, I believe adding TCP is a good thing, and that
> to do otherwise "hints" of practicing a technology religion
> as opposed to heeding market needs. I also think TCP is
> extremely useful in fault management applications as that
> helps SNMP managers implement event driven alarm models instead
> of poll driven models. I gather poll driven alarm models are
> considered next to useless in at least many (most?) telecom arenas.
I doubt that we must use TCP for the event drive model.
We now see a lot of managers do "trap directed" polling. In other
words they go around all agents every 15 (or whaever) minutes
to make sure the agent is still there. If a trap is received we
quickly check the generating agent.
But we could define a mechanism by which a agent would send
"hartbeat traps" every so often (15 minutes or whatever) and
one of the objects in the trap would be a "heartbeat counter".
If a manager does not get the heartbeat trap when expected
it could check out the agent. If it does get the heartbeat it can
check from the counter if it missed any.
This way, we can stick to UDP, we only have one way traffic
(as opposed to two-way traffic).
The idea (in fact part from a larger idea) comes from Geoff
Carpenter several years ago when he was still at IBM.
Apology for using mgr/agent terminology ... :-)