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

20021114: Questions about NCAR application



>From: "Peter O'Neil" <address@hidden>
>Organization: UCAR
>Keywords: 200211142212.gAEMCPL03300 IDD

Charles,

Peter O'Neil of NCAR forwarded an email exchange that you had with Marla
Meehl concerning your use of the Unidata LDM for moving data to your
institution.

Since the latencies indicated by traceroutes to your site (ust.hk) are
about the same as from Boulder to a site that we have had great success
in providing data in Brazil, we feel that our LDM should provide you
with reliable data delivery.  It may be the case that some simple tuning
of your request for data to your upstream feed site(s) could result
in elimination of problems you have apparently been having.   In order
for us to advise you on what can be done, we will need several pieces
of information:

o the fully qualified hostname of the machine you are running the LDM
  on

o the version of the LDM you are running

o the fully qualified name of the upstream host that is feeding your
  LDM data

o what datastreams you are trying to feed

o copies of your ldmd.conf and pqact.conf files

o what operating system (vendor, release, etc.) you are running on
  your LDM machine

I look forward to hearing from you in the near future.

>--On Tuesday, November 12, 2002 12:55 PM -0700 Marla Meehl <address@hidden> 
>wrote:
>
>> David and Peter,
>>   Here is more information on the Hong Kong application.
>>
>> Marla
>>
>> Charles Choy wrote:
>>>
>>> Hello Marla,
>>>
>>> The HKUST CCAR department regularly retrieves data
>>> from NCAR using the LDM protocol.  On and off,
>>> we have time-out problems and the data retrieved is
>>> not completed.
>>>
>>> We have studied the LDM protocol definition which
>>> mentions that the network propagation delay should
>>> be at most 150ms.  Unfortunately, due to geographical
>>> distance between Hong Kong and US, the minimal network
>>> delay we can achieve is 240ms to 280ms.
>>>
>>> FYI, we have a direct DS3 link to Abilene and so our
>>> bandwidth should be abundant.
>>>
>>> We would be most grateful if you can provide hint on
>>> tackling this problem.
>>>
>>> Thanks
>>> Charles
>>> HKUST
>>>
>>> > -----Original Message-----
>>> > From: Matthew E. Schmitz [mailto:address@hidden]
>>> > Sent: Tuesday, November 12, 2002 9:33 AM
>>> > To: address@hidden
>>> > Cc: address@hidden; address@hidden; address@hidden
>>> > Subject: Questions about NCAR application
>>> >
>>> >
>>> > Hello Marla,
>>> >
>>> > I hope all is well with you in CO.  I head you guys had quite
>>> > the Halloween in terms of weather.  Sounds pretty much polar
>>> > opposite to hot and humid Singapore.
>>> >
>>> > I wanted to introduce to you some folks I met the other week
>>> > at HKUST (http://www.ust.hk/en/aa/index.html).  Campus was a
>>> > Hong Kong version of Boulder (way up in the hills).  Mr. Law,
>>> > Mr. Wong, and Mr. Choy have been working with some
>>> > researchers on some atmosphere data base application (based
>>> > at NCAR) that has some strict latency requirements.  Given
>>> > the distance and link bandwidth to Hong Kong, Mr. Law, Mr.
>>> > Wong, and Mr. Choy were wondering if there was any one they
>>> > could talk to at NCAR about application tuning and tweaking.
>>> > I don't have exact details but figured they could tell you more.
>>> >
>>> > Thank you for any assistance you can lend your counterparts
>>> > here in Hong Kong.
>>> >
>>> > Regards,
>>> > Matt

Tom Yoksas