Current Status of NOAAport

by Robb Kambic   
<p>

On December 17, 1998 the UPC changed from the WMO data stream to the NOAAport 
data stream.  The change required the cooperation of all the sites in the IDD
network because the stream size, the stream data rate, the network congestion, 
new LDM software configurations, increased CPU load, and new topologies. There 
was a test period to expose potential problems before the final switch.  The 
transition went smoothly and the new data products are an interest to all. 


The NOAAport stream is a super-set of the FOS stream.  The differences include 
the number and types of products as well as the size of products. This caused 
some sites network links to become saturated that resulted in the sites 
requesting subsets of the data streams. On the other hand, the vBNS(high speed 
links) connected sites can handle the full data streams and distribute to other 
vBNS sites in less than a minute on average.  With minor rearrangement of the 
IDD topology most sites are satisfied with NOAAport, especially since many of 
the Model data arrives as much as three hours earlier than the older satellite 
systems. The sites are especially interested in the new Models and graphical 
products on NOAAport, most administrators are still in the progress of 
decoding/using the data. This activity has sparked much e-mail though-out the 
IDD community. The FOS data size was about 20 megabytes and the NOAAport data 
stream can peak at 110 Megabytes with and average around 50 megabytes per hours.
The increased volume caused some machines to become CPU bound causing hardware 
upgrades to happen at an increased rate. The comments from our community of 
users has been positive even with the increased burden on the remote sites 
networks and machines. This change has also a monetary benefit of releasing 
the cost to access the FOS data.  But the implementation of NOAAport does come 
with a cost of satellite hardware and receivers.  But, the is no charge for 
the data per se. The present situation is positive with out looks that the 
sites only receiving a subset of the data will be able to receive the needed 
data via network improvements and cheap computers in the near future.


The source data distribution and the redundancy was also changed because of 
NOAAport. There is a ring of source sites that all have NOAAport satellite
dishes. The architecture for the source sites is to feed each other so if
a NOAAport satellite dish fails at any one site then they will get the data
from another site.  The source sites are UCAR, SSEC, and Alden. When other
sites receive NOAAport dishes they will also be added to the ring of source
sites. UCAR also has another backup machine called iita.rap.ucar.edu to the
main IDD distribution machine thelma.ucar.edu. The was mainly installed because
of potential hardware failures on thelma. 

The fail over scheme is:

- if thelma fails, fall over to iita.  This will result in the minimal
  network changes.
- If both thelma and iita are unreachable, fail over to SSEC and Alden. The 
  sites will have to make the decision for the best connection.
- There will also be an alternative site, Cornell for the northeast. Cornell
  is also positioned on the vBNS.
 

This initial reactions to NOAAport has been very positive. It has caused an
increased number of support questions and rearrangement of a few IDD sites.
This NOAAport change is temporary until the LDM can be modified to handle 
NOAAport feedtypes and the original NOAAport headers. Currently the NOAAport
headers are used to create WMO headers, this results a loss in some of the
NOAAport header information. These changes are expected to be completed in the
second quarter at the latest. 

Robb Kambic