[32901] in North American Network Operators' Group
Re: Alter.net and Verio peering in Sacramento?
daemon@ATHENA.MIT.EDU (John Fraizer)
Sun Dec 17 15:21:29 2000
Date: Sun, 17 Dec 2000 15:19:01 -0500 (EST)
From: John Fraizer <nanog@EnterZone.Net>
To: nanog@rmrf.net
Cc: "Christopher L. Morrow" <cmorrow@UU.NET>,
Randy Bush <randy@psg.com>, Mathew Butler <mbutler@tonbu.com>,
"'nanog@merit.edu'" <nanog@merit.edu>
In-Reply-To: <Pine.LNX.4.21.0012162308020.12826-100000@zeus.its-my.net>
Message-ID: <Pine.LNX.4.21.0012171510340.24487-100000@Overkill.EnterZone.Net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Errors-To: owner-nanog-outgoing@merit.edu
On Sat, 16 Dec 2000 nanog@rmrf.net wrote:
> We at SAVVIS run continous pings to every backbone router that we
> own. (John - this does not include our ATM switches, we have no way to
> monitor latency between them). But if, say, our connection to Sprint or
Um, sure you do. You know your atm mesh and (hopefully) you have an IP
address bound up on the switch itself so you can gather snmp statistics
from the box.
I realize that this might have been too obvious to have come up in
discussions when proactive monitoring was being implemented on your
net.
If you know that from your monitoring station it's normally 6ms to switch
X and switch Y lives (normally) 6ms behind switch X, it's simple to set up
alarm profiles.
if X > 6 alarm
if Y > 12 alarm
You get the general idea. We do this exact same thing in our nework and
have dependencies set up as well. For example, switch monitoring is
dependent on switch X not being in alarm so, if X is in alarm, don't alarm
for Y also because our path to Y is through X. blah blah.
Don't worry about the CBX-500 in CMH. You know we'll call if it has
problems. <grin>
---
John Fraizer
EnterZone, Inc
