[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: e2e measurements



On end-to-end measurement...  This topic is broader than the goals of
psamp (and arguably outside of psamp's scope), though it is worthwhile
to note that packet sampling could be used to supplement e2e active
measurements.  For example, suppose active measurements are sent over
a path that traverses a psamp-enabled link.  The link could be
configured with to filter on the IP addresses of the probe machines
and sample these packets aggressively (say, at 100%).  This could give
a view of how the probe traffic looks inside the network (say, to
study the loss or jitter on the portions of the path from the source
probe machine to the psamp link, and from the psamp link to the
destination probe machine).

> Is their any research on e2e measurements.

There is a large body of work on using e2e measurements to infer
link-level properties, such as loss rates, capacity, available
bandwidth, etc., and to characterize path-level properties of the
Internet.  Here's an (incomplete) list...

  - ftp://ftp.ee.lbl.gov/papers/vp-pkt-dyn-ton99.ps.gz
  - ftp://ftp.ee.lbl.gov/papers/vp-routing-TON.ps.Z
  - http://www.employees.org/~bmah/Software/pchar/
  - http://www.acm.org/sigcomm/sigcomm99/papers/session7-1.html
  - http://www-net.cs.umass.edu/minc/
  - ftp://ftp-sop.inria.fr/rodeo/bolot/93.Internet.ps.gz 
  
There is also work on e2e characterization of Web performance.  See

  - http://www.icir.org/vern/imw-2001/imw2001-papers/10.pdf
  - http://www.cs.bu.edu/techreports/1999-004-wide-area-web-measurement.ps.Z
  - http://www.research.att.com/~bala/papers/www9.html

> Are the selection of internet performance metrics done from end-user
> perspective

Some relevant IETF activity includes

  http://www.ietf.org/html.charters/ippm-charter.html

> Does netscope take end-user persecptive while selecting a metric?

By "netscope" are you referring to the tool developed at AT&T Labs for
traffic engineering?  I'm not sure exactly how this is relevant to the
discussion, since NetScope does not conduct measurements but rather is
used to display measurements, and to overlay traffic measurements on
topology and routing information.  By "selecting a metric" I assume
you mean "selecting an optimization criteon for routing."  This metric
could be based on network load, user performance, etc., though it is
arguably difficult to select metrics that are fine-grain with regard
to user performance (such as throughput of individual TCP connections)
without very fine-grain measurement data and good models of how user
sessions would behave after a change in routing.

-- Jen

--
to unsubscribe send a message to psamp-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/psamp/>