From: Luis F Balbinot (hades@inf.ufrgs.br)
Date: Mon Jun 18 2001 - 06:21:08 PDT
On Sun, 17 Jun 2001, David Cheney wrote:
DC>What analyses (if any) do you use? What do you use them for? If
DC>these analyses are useful only for selecting traces to download,
DC>then how well do they suit your needs? What (general) analyses would
DC>you like to see?
Well, post-processing of traces is a very relevant issue when we are
dealing with large trace files. Sometimes these traces are huge (see
Waikato traces) and are just impossible to download through the Internet
(worldwide). But instead of having general scripts to post-process these
traces, what about allowing the users to send their own scripts to make
local processing of these traces?
There are several issues with that too. Security, usage of resources, etc,
but there are some standardized ways to do that.
About 1,5 years ago we started to deploy some CoralReef-based measurement
stations in our local Internet2 links here in southern Brazil. We faced
the same problems with the traces. I wrote a proposal to make the
distributed analysis of traces, allowing users to delegate scripts (Perl)
using the SNMP framework.
If there is anyone interested in this research, the paper is available for
download at http://cscience.org/~hades/papers/ipom2000.ps.gz. It is not a
"state of the art" paper, but I guess the ideas are valid.
I also brought this issue up because we are starting to deploy passive
measurement probes at the Brazilian Research Network, and this issue will
arrise (again) for us soon.
-- Luis F Balbinot hades@inf.ufrgs.br http://cscience.org/~hades/
This archive was generated by hypermail 2b30 : Thu Sep 27 2001 - 16:24:41 PDT