[Planetlab-users] CoMon archived data

[Vivek Pai]

Chuang Liu wrote:
> Hi:
> I would like to access the archived data collected by CoMon in June (or 
> any other month). Can anyone give me some instructions about how to do 
> this?
> 
> I followed instructions in http://comon.cs.princeton.edu, and it doesnot 
> work. The URL I used to access June 1 data is
> http://comon.cs.princeton.edu/status/dump_comon_20050601.bz2
> Any hints?

Well, there's this line on the web page:

    "Older data may be moved to another directory or offline -- if you
     need it, please contact us."

or this posting to the users list 11 days ago:

-------- Original Message --------
Subject: Re: [Planetlab-users] The fate of All Pairs Pings
Date: Sun, 03 Jul 2005 13:32:11 -0400
From: Vivek Pai <vivek at CS.Princeton.EDU>
To: Hakim Weatherspoon <hweather at cs.berkeley.edu>
CC: users at lists.planet-lab.org
References: <42BC2784.6090601 at pdos.lcs.mit.edu> 
<Pine.GSO.4.61.0507021241530.9090 at argus.EECS.Berkeley.EDU>

Older CoMon data's available at
http://comon.cs.princeton.edu/status/dumps_old/
It goes back to August 2004, but it's a little spotty due to hardware
failures.

Hakim Weatherspoon wrote:
 > On Fri, 24 Jun 2005, Sean C. Rhea wrote:
 >
 >> On Jun 24, 2005, at 8:32 AM, Jeremy Stribling wrote:
 >>
 >>> 1) Make it an all sites ping: only measure pairwise latency and
 >>> availability in terms of sites (run all pairs pings among one node per
 >>> site).
 >>
 >>
 >> The all-sites ping idea would work just as well for that sort of
 >> problem, and we could do all-pairs-ping within sites for availability
 >> numbers if we want.
 >>
 >>> 3) Discontinue it forever, and use services like
 >>> http://monitor.planet-lab.org and http://comon.cs.princeton.edu for
 >>> availability information.  I don't believe these sites are archived,
 >>> but that should be easy for someone to implement.
 >>
 >
 > I have not seen much discussion on this topic, but I would vote for
 > option 1) above and would be willing to help.  Also, option 3) above is
 > a viable option if an archival repository of the data were made 
available.
 >
 > I have been using all-pairs ping data to create historical availability
 > traces that I use in simulation.  The all-pairs ping data covers nearly
 > all of planetlabs history (2 1/2 years from feb 2003 to current) and is
 > nearly the only archival repository that does. But I guess its time has
 > come to let a more scalable technology be used as the defacto standard
 > to monitor planetlab.  My strongest request is that this replacement
 > (e.g. option 1 or 3 above) be archived continuously.  I am willing to
 > help in this process.
 >
 > -Hakim
 >
 > _______________________________________________________________
 >   Hakim Weatherspoon hweather at cs.berkeley.edu
 >   http://www.cs.berkeley.edu/~hweather
 >
 > On Fri, 24 Jun 2005, Jeremy Stribling wrote:
 >
 >> Hi everyone,
 >>
 >> It appears that the All Pairs Pings service
 >> (http://pdos.csail.mit.edu/~strib/pl_app) is reaching its scalability
 >> limits. As the name implies, the service can't exactly scale forever,
 >> and with clusters of tens of nodes coming up in Japan and China, the
 >> bandwidth is becoming unmanageable.  So PlanetLab Central and I have
 >> decided it might a good time to shut it down soon.
 >>
 >> I would like to hear users' thoughts on this.  I have no idea how many
 >> people use it, or how useful people think it is.  I will keep the
 >> historical data on the website for as long as I stay at MIT (another
 >> 3-4 years at least), but as far as the continuous updates go, there
 >> are a few options:
 >>
 >> 1) Make it an all sites ping: only measure pairwise latency and
 >> availability in terms of sites (run all pairs pings among one node per
 >> site).
 >> 2) Bring the period of pings waaaaay down (say once a day).
 >> 3) Discontinue it forever, and use services like
 >> http://monitor.planet-lab.org and http://comon.cs.princeton.edu for
 >> availability information.  I don't believe these sites are archived,
 >> but that should be easy for someone to implement.
 >>
 >> Either way, I think I want to lessen my involvement in the project,
 >> and if people want the service continued in some other form I'd like
 >> some help.  I'm happy to make ll the code available to anyone who is
 >> interested.
 >>
 >> Let me know what you think.  Thanks,
 >>
 >> Jeremy
 >>
 >> _______________________________________________
 >> Users mailing list: Users at lists.planet-lab.org
 >> https://lists.planet-lab.org/mailman/listinfo/users
 >>
 >
 > _______________________________________________
 > Users mailing list: Users at lists.planet-lab.org
 > https://lists.planet-lab.org/mailman/listinfo/users
 >


_______________________________________________
Users mailing list: Users at lists.planet-lab.org
https://lists.planet-lab.org/mailman/listinfo/users