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

20040203: test feed from emo...feedback requested (cont.)



>From: Mike Voss <address@hidden>
>Organization: SJSU
>Keywords: 200402040240.i142exp1018335 IDD test

Hi Mike,

>Well, storm.sjsu.edu has been feeding from emo for a few days now and hasn't
>skipped a beat, receiving all the data with limited latencies. Is this how
>you read the stats also?

Yes, exactly.

>Just for kicks I just ftp'd some 1/2 GB files from motherlode and I was able
>to get 12.8 Mbps throughput on the transfer, while the LDM did not skip a
>beat:
>----------------
>ftp> get 2004020318_eta212.gem
>local: 2004020318_eta212.gem remote: 2004020318_eta212.gem
>227 Entering Passive Mode (128,117,13,119,75,178)
>150 Opening BINARY mode data connection for 2004020318_eta212.gem (440915968 b
> ytes).
>226 Transfer complete.
>440915968 bytes received in 269 secs (1.6e+03 Kbytes/sec)
>-----------------

Looking good.

>This is a huge breakthrough here, It seems I've waited years for this.

There is nothing like a fast Internet connection.

>I will wait to move storm.sjsu.edu off of emo until you think we have enough
>statistics to say it's fine.

I think that the message is clear at this point.  All you have to
do is show the comparison of HDS latencies on rossby:

http://my.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?HDS+rossby.met.sjsu.edu

with the same on storm:

http://my.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?HDS+storm.sjsu.edu

Then throw in the plot of total volumes received by each:

rossby:
http://my.unidata.ucar.edu/cgi-bin/rtstats/rtstats_summary_volume?rossby.met.sjsu.edu+GRAPH

storm:
http://my.unidata.ucar.edu/cgi-bin/rtstats/rtstats_summary_volume?storm.sjsu.edu+GRAPH

Storm is getting over 4 times the amount of data, and its latencies are
in the several seconds range for all feeds.  Rossby's latencies for
HDS, on the other hand, range up to over 1000 seconds at the same time.
If this is not a clear indication of throttling, nothing is.

>After that I'm going rename storm.sjsu.edu to
>methost24.met.sjsu.edu, move it inside the College of Science network, and
>try to feed everything from emo again. This should be the final proof that
>we have a major network limitation in our College.

Sounds good.  Make sure to grab some plots before you move storm
around.

>Let me know if you have any feedback, thanks,

See above.

Cheers,

Tom
--
NOTE: All email exchanges with Unidata User Support are recorded in the
Unidata inquiry tracking system and then made publically available
through the web.  If you do not want to have your interactions made
available in this way, you must let us know in each email you send to us.