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

Re: Multi-feeds into a decoder (dcuair)

On Tue, 7 Dec 1999, Gregory Grosshans wrote:

> The timeliness of upper air data via AFOS has always been faster than
> FOS (DDS).  Now  NWS sites are gradually moving to a state where AWIPS is
> 'Operational' and they transmit all products on AWIPS.  These products are
> eventually routed to the FOS, but not to AFOS (acceptable).  Thus, the data
> is still slow to arrive via FOS, but quick for those with NOAAPORT.
> 
> For several years LDM and pqact.conf have been configured with the following
> at the SPC:
> 
> AFOS|DDS        ^...(MAN|SGL)...|^U[EGHKLPQSZ].... .... ([0-3][0-9])
>         PIPE    dcuair -v 2 -d logs/dcuair_combines.log -m 24
>                 data/decoders/uair/YYMMDD.snd
> 
> This has worked well and allowed the forecasters to view the UA data in a
> timely fashion (since the timely UA data is actually from AFOS).
> 
> I have NOAAPORT data feeding into LDM with a stream type 'FSL5'.  Has anyone
> ever configured LDM with three separate streams flowing into one decoder?

Greg,

The architecture of the LDM is queue based, so the different streams are
entered into the queue.  The pqact program is a queue reader, doesn;t care
how the products are entered into the queue. The pqact program scans the
queue, get the product ID's and then sequentially going through the pqact
entries to see if there is a match.  If a match is found then the entry is
invoked on behave of that product. 

> Are there any serious performance problems with doing this? 

Since the products are coming from different streams, there probably
will be no duplicate detection provided by the LDM to remove duplicates.
So all the products will be sent to your decoder from the different
streams. The question is: Can your decoder handle the duplicates?

Robb...

 I'm considering
> setting up something similar to the following:
> 
> AFOS|DDS|FSL5        ^...(MAN|SGL)...|^U[EGHKLPQSZ].... .... ([0-3][0-9])
>         PIPE    dcuair -v 2 -d logs/dcuair_combines.log -m 24
>                 data/decoders/uair/YYMMDD.snd
> 
> The two or three streams provide redundancy, but more importantly provide the
> data to the forecaster ASAP.
> 
> 
> Thanks,
> Gregg Grosshans
> Storm Prediction Center
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================