Hi All,
Rich: Does this in any way affect the Uptime % ? (and is that now
remembered between server restarts? :-/ )
If it in any way affect the uptime %, then I would argue for the old
model, but if not, then I see benefits in the newer model.
One being, packet loss before we completely lost the connection,
could help indicate it was a problem due to cable or other hardware
issues, (+1% loss) or if the loss is zero, it might be something else.
Should it apply to Short & Long term?
Hmm, I would say yes, if the probe still indicate that it has lost N
packets since it went into Down state.
PS: If I'm not clear, it might be because I only got 2 hours sleep
after the Madonna concert last night. 85.000 people here in DK...
what a gig!
Jakob Peterhänsel
"Be a part of the Love Generation - carry a smile, not a gun."
- JP, May 2006
Email: [EMAIL PROTECTED]
AIM: Marook
Phone: +45 22684961
On 25/08/2006, at 5:39, Richard E. Brown wrote:
Folks:
As you're aware, InterMapper sends ping and SNMP packets, and uses
the fraction
of returned packets to compute packet loss statistics. We're
looking for some
advice about the algorithm:
1) InterMapper 4.5 introduces a Short-term Packet Loss statistic.
InterMapper
remembers the history of the last 100 packets. The fraction of lost
packets is
displayed in the device's Status Window.
2) Version 4.5 also implements a new way to compute packet loss.
Here's a
description of the current way (4.4) and then the new (4.5).
Earlier versions of InterMapper counted the number of pings and
snmp queries
sent, and the number of pings/snmp responses received. It did the
simple
calculation with these values to get percentage loss.
InterMapper 4.5 does essentially the same thing, except that it it
does not
update the percent loss statistics when the device is down. That
is, after a
sufficient number of lost packets (default is 3), the packet loss
statistics are
not updated until InterMapper hears a response from the device.
The argument for the old behavior is that it's simple to explain.
The difficulty
with that behavior is that an outage will cause the reported error
rate to creep
up as dropped packets accumulate.
The behavior in version 4.5 attempts to preserve the packet loss
stats from
before the outage, so that a network administrator can determine
what the packet
loss was before the failure.
The questions to the InterMapper-Talk list are:
- Which behavior is more useful?
- Are there circumstances where ignoring lost packets during
outages would give
a "wrong" result?
- Should that behavior apply to both short-term and long-term
packet loss?
Many thanks!
Rich Brown [EMAIL PROTECTED]
Dartware, LLC http://www.dartware.com
10 Buck Road, PO Box 130 Telephone: 603-643-9600
Hanover, NH 03755-0130 USA Fax: 603-643-2289
____________________________________________________________________
List archives:
http://www.mail-archive.com/intermapper-talk%40list.dartware.com/
To unsubscribe: send email to: [EMAIL PROTECTED]
____________________________________________________________________
List archives:
http://www.mail-archive.com/intermapper-talk%40list.dartware.com/
To unsubscribe: send email to: [EMAIL PROTECTED]
