[ANSS-netops] Basalt data flow and bandwidth limitations

Jon Rusho jon at seis.utah.edu
Tue Oct 25 17:41:42 UTC 2011


All,

  For those of you that have installed Basalts as part of ARRA and are 
using bandwidth-limited connections (e.g., spread-spectrum radios), you 
might be interested in this.

UUSS had been using the RockToEW module on the Basalts to export data 
into our Earthworm systems.   We'd noticed a lot of gaps in our data 
flow on radio networks but not on other telemetry means (e.g., cell 
modem, dsl, etc).   When looking at our cell modem usage, a Basalt 
streaming 4 channels at 100sps was using almost twice the estimated 
bandwidth.   It turns out that there is a lot of packet overhead and no 
compression of the data.  

While looking into the problem, we tested the RockToSLink module for the 
Basalts, which can send data to SEEDLink-client programs.   This module 
supports STEIM2 compression.   In testing, this showed a significant 
reduction in network traffic (50% or better).   An additional benefit to 
the RockToSLink module is that it maintains a buffer and if telemetry 
drops, upon reconnect the client can request older data.   
So far, we've changed all but three of our Basalts to the RockToSLink 
module and are using the Earthworm slink2ew module to bring in the 
data.   For a pair of sites we could never get more than a few minutes 
of continuous data before, the data has been virtually gapless (a couple 
of gaps were introduced by human error).
There are a couple of tradeoffs such as a slight increase in latency and 
a change in the tracebuf2 size once the data is in Earthworm.   The 
change in the tracebuf2 size is related to the SEEDLink data exchange 
and required some minor changes to our wave_serrverV configuration.   
 From our experience so far, the continuous data is well worth the extra 
work and slight latency.

So...if you're having telemetry problems with Basalts and are using 
RockToEW, we suggest you try RockToSLink instead. 

Both RockToEW and RockToSLink are modules created by ISTI to run on the 
Rock platform from Kinemetrics.


-- 
===================================
Jon Rusho
Seismic Network Engineer 
University of Utah Seismograph Stations

jon at seis.utah.edu
801-585-5523




More information about the ANSS-netops mailing list