NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
As an experiment, Unidata is currently relaying high resolution model output from Boulder to Washington, D.C. via INN. This is a high volume stream ranging from 23 to 25 gigabytes of data per day. A test starting on January 11 and extending for over 24 hours resulted in the successful relay of over one half million products with 99% arriving within 4 seconds or less:
Product Latencies
Total values read: 549622
secs #prods % cumulative%
---- ------ ---- ------------
-1: 1572 0.29 0.29
0: 364983 66.41 66.69
1: 149892 27.27 93.96
2: 21588 3.93 97.89
3: 5440 0.99 98.88
4: 913 0.17 99.05
5: 413 0.08 99.12
10: 652 0.12 99.24
20: 1699 0.31 99.55
30: 1416 0.26 99.81
40: 975 0.18 99.99
50: 79 0.01 100.00
Graphs showing current 5 minute statistics are available "NNTP Relay Stats". These four graphs show (1) average latency per second, (2) products received per second, (3) minimum latency per second, and (4) maximum latency per second.
Products relayed via Usenet for this test are under the group unidata.*
with a 'Distribution: unidata' header. Currently all unidata.binaries.*
articles will be sourced with a path header entry of imogene.unidata.ucar.edu.
Note that to receive a data product via NNTP and view it, the product would need to be decoded from its transmission format and then subsequently decoded into a format compatible with the chosen visualization package.