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

20030516: request to upgrade to latest Unidata LDM release, 6.0.11 (cont.)

>From: Brendon Hoch <address@hidden>
>Organization: Lamont Doherty
>Keywords: 200305161525.h4GFPILd019927

Hi Brendon,

I truely appreciate the comments you include below, but I want to take
this opportunity to try and convince you to upgrade to LDM-6 :-)  Please
accept my comments in that spirit.

>I've been hesitant to install the LDM 6 because of the numerous messages
>received on the mailing lists from bleeding edge users.

Most every inquiry related to use of current LDM-6 releases (6.0.10 and
6.0.11) have been traced back to user issues.  The LDM-6.0.10/11 is the
most stable and efficient release of the LDM that we have ever made.
It is this version that is able to relay all of the high resolution
model data produced by NCEP (the CONDUIT stream) throughout the country
with very small latencies.  It is also this version that will be used
to relay the NEXRAD Level II radar data by the NWS.

>According to
>your website, there have been 3 minor releases to resolve many of the
>problems between March 12th and May 13th.

That is correct.  The minor releases have been, in fact, very minor:
the change from 6.0.10 to 6.0.11 was a one liner.  That mod changes the
tenacity with which a receiving rpc.ldmd will attempt to reconnect to
an upstream feed site that has apparently disappeared for an extended
period of time.

>Our usage of LDM is somewhat
>mission critical, thus I have waited to upgrade until all of the bugs
>are worked out.  The 6.0.11 release has only been out for less than a
>week, so it would be nice to give the bleeding edge users a chance to
>give this a test drive.

From our standpoint there is essentially no oerational difference
between 6.0.10 and 6.0.11.

LDM-6.0.10 has been in continuous and heavy use at all
Unidata-controlled IDD machines (including _all_ top level nodes) for
about a month and a half.  Stress tests on the most loaded IDD relay
machine, thelma.ucar.edu, has shown that the LDM-6 was capable of moving
over 100 Mbps of data in burst mode and 55 Mbps on average without
introducing latencies in any of the feeds being relayed.

Some examples of the dramatic improvements in data relay (from a
latency point of view) can be seen in:

Example of latency drop when a site's upstream upgraded to LDM-6:

http://www.unidata.ucar.edu/staff/tom/gifs/LDM-5_to_LDM-6.gif

Example of drop in latencies when the NWS CONDUIT injection machine
was upgraded to LDM-6:

http://my.unidata.ucar.edu/content/committees/polcom/pcmeeting2003May/NCEP_CONDUIT_Motherlode.gif

>The fact that 6.0.11 is only available in
>source form makes me suspicious that there may be 1-2 additional minor
>releases to come out in the near future.

We will, of course, continue to make small changes when and where
warranted, but this has nothing to do with the availability of binary
releases.  During the rewrite of LDM, we took a close look at how
binary releases were installed and did not like what we saw.  Quite
frankly, some of us are in favor of only making source releases.

>I do intend to upgrade, but
>only when the commotion dies down a bit.

Since your use of the LDM is mission critical, I want to again
encourage you to upgrade to LDM-6 as soon as you can.  I am confident
that you will find that receipt and possible relaying of data will
be much faster and more reliable than with LDM-5.

>_______________________________________________________
>Brendon Hoch
>International Research Institute for Climate Prediction
>Earth Institute at Columbia University
>141/208 Monell Bldg., Lamont Doherty Earth Observatory
>61 Route 9W, PO Box 1000, Palisades, NY 10964
>-------------------------------------------------------
>Phone: (845)680-4444 Fax:(845)680-4488
>mailto:address@hidden
>http://iri.columbia.edu/~bhoch
>-------------------------------------------------------

Thanks for getting back to us on this!

Cheers,

Tom Yoksas