Re: tcptrace Feature Request. Suggestion

From: Jose Manuel Cano Garcia (cano@pc2072te.dte.uma.es)
Date: 09/16/02

  • Next message: Mark Allman: "Re: tcptrace Feature Request"

    From: Jose Manuel Cano Garcia <cano@pc2072te.dte.uma.es>
    Subject: Re: tcptrace Feature Request. Suggestion
    Date: Mon, 16 Sep 2002 23:07:15 +0200
    Message-Id: <200209162307.15056.cano@pc2072te.dte.uma.es>
    
    

    Hi,

            I have one suggestion to the csv output format. The real important question
    is not the order, but the delimiters of the fields. I am interested in
    loading traffic traces in matlab, and it could only be accomplished easily
    (by means of the matlab's "load" command) if the fields are delimited by
    blank spaces or tabs instead of commas. So I suggest to use a configurable
    delimiter,if possible (of course, "sed" may be used instead...).

    Finally, let me make a different and not related suggestion:

            Sometimes I have to process asymmetric traces, in which packets corresponding
    to one of the transmission senses are lost. However, the amount of data
    transmited in this "invisible" link, could be estimated using the ACK
    sequence numbers of packets on the reverse link. ¿May it be possible to
    modify tcptrace to track the amount of acked data?

    Thank you very much, tcptrace is really a very useful program.

            Best regards,

                    Jose

    =====================================
    Jose Manuel Cano-Garcia
    Departamento de Tecnologia Electronica
    Universidad de Malaga, ETSI Telecomunicación
    Malaga - Spain
    ====================================

    El Lun 16 Sep 2002 19:34, Kevin escribió:
    > In truth I can live with either. Given the choice, the 2nd option is
    > what I am building my current awk/sed script to. It just seems simpler
    > to hand off pairs of data to be counted. I also suggest the
    > source/dest (a/b) as the 1st field of the line.
    >
    > Thanks for the chance to have input!
    > Kevin Mason
    >
    > > -----Original Message-----
    > > From: alakhian@masaka.cs.ohiou.edu
    > > [mailto:alakhian@masaka.cs.ohiou.edu]
    > > Sent: Monday, September 16, 2002 13:28
    > > To: Kevin
    > > Cc: tcptrace@tcptrace.org
    > > Subject: Re: tcptrace Feature Request
    > >
    > >
    > > Kevin,
    > >
    > > We discussed about your feature request and believe that it
    > > would indeed be a good idea, and very flexible too as Mark
    > > said. The changes to output.c would be very trivial and we
    > > could add a --csv option. The question remains as to what
    > > would be the best representation of the data fields so as to
    > > make this option really useful. I gave this some thought but
    > > I believe that since you and many other would be the ones to
    > > actually use this feature, it would be better to get some
    > > feedback first.
    > >
    > > Since you would want to use this data for some sort of
    > > further processing, obviously the text would not be very
    > > useful. So, would it be useful to output all the data in the
    > > same order (as the long output) separated by commas with a->b
    > > first followed by b->a. Or should the data be sorted based on
    > > fields such as:
    > >
    > > total packets: 52 total packets: 38
    > > ack pkts sent: 51 ack pkts sent: 38
    > >
    > > to look like 52,38,51,38
    > >
    > > where the data is ordered as total_packets_ab,
    > > total_packets_ba, ack_pkts_sent_ab, ack_pkts_sent_ba ...
    > >
    > > Comments/Suggestions?
    > >
    > > Thanks!
    > >
    > > --Avinash
    > > (tcptrace-maintainer)
    > >
    > > On Mon, Sep 09, 2002 at 11:36:43AM -0400, Kevin wrote:
    > > > My apologies if this is being sent to the wrong list.
    > > >
    > > > When using tcptrace -l a lot of very useful information is output.
    > > > When there are lots of sessions (>20) the output format is
    > >
    > > cumbersome
    > >
    > > > to compare the various sessions.
    > > >
    > > > Would it be possible to have to have an option to output in CSV
    > > > format? That way the data can be sorted and sifted to
    > >
    > > compare various
    > >
    > > > errors or performance values.
    > > >
    > > > Thanks
    > > > Kevin Mason
    > >
    > > --
    > > Avinash Lakhiani (http://www.tcptrace.org/~alakhian)
    > > --
    > >
    > > ---
    > > Incoming mail is certified Virus Free.
    > > Checked by AVG anti-virus system (http://www.grisoft.com).
    > > Version: 6.0.386 / Virus Database: 218 - Release Date: 9/9/2002
    >
    > ---------------------------------------------------------------------------
    >- To unsubscribe, send a message with body containing "unsubscribe tcptrace"
    > to majordomo@tcptrace.org.

    ----------------------------------------------------------------------------
    To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
    majordomo@tcptrace.org.



    This archive was generated by hypermail 2b30 : 09/17/02 EDT