[174133] in North American Network Operators' Group
Re: network quality measurement probes+reporting
daemon@ATHENA.MIT.EDU (Joe Loiacono)
Tue Aug 26 14:02:50 2014
X-Original-To: nanog@nanog.org
In-Reply-To: <20140826171314.GA11793@pob.ytti.fi>
To: Saku Ytti <saku@ytti.fi>
From: Joe Loiacono <jloiacon@csc.com>
Date: Tue, 26 Aug 2014 14:02:30 -0400
Cc: nanog@nanog.org, NANOG <nanog-bounces@nanog.org>
Errors-To: nanog-bounces@nanog.org
Perhaps you've considered this already ... not sure if it gets 1us or not
: bwctl?
http://software.internet2.edu/bwctl/
Joe
From: Saku Ytti <saku@ytti.fi>
To: nanog@nanog.org
Date: 08/26/2014 01:15 PM
Subject: network quality measurement probes+reporting
Sent by: "NANOG" <nanog-bounces@nanog.org>
Anyone can recommend or even just name drop network quality measurement
kit?
I'm only familiar with IP SLA, RPM and Creanord (and various inhouse
tools:)
What I'd like to see
- 1us or better one-way jitter (no need for clock sync, just accurate
clock)
- tens or 100us one-way latency (as good as cheaply can get sync, ntp is
ok)
- 1us or better RTT
- any-to-any measurement, not just hub<->spoke (or sufficiently cheap
hub)
- 100pps to 100 measurement points on 8 CoS (ish, less may be
acceptable)
- randomized payload pattern + verification (to catch bit mangling)
- randomized sport/dport (to put traffic in each ECMP/LAG combo, long
term)
- programmatic accesss and useful documents to measurement data (e.g.
some OSS TSDB)
- high quality, fast, useful graphical reporting and alarming
- support for TWAMP and OWAMP responders
- indicative price <100kEUR CAPEX and <2000EUR YRC for 100 nodes
solution
--
++ytti
