Is ROSS Ready forBy Larry Dignan | Posted 2003-12-04 Email Print
California uses a real-time system to rally resources to fight wildfires. Will it hook up with a national system?Primetime?"> The biggest question for the California fire defense team: Is ROSS ready for primetime? MIRPS project manager John King says it's unlikely that the CDF will use ROSS next fire season, but 2005 is a possibility.
ROSS was developed by Lockheed Martin for the National Wildfire Coordination Group and designed to accommodate multiple federal agencies and all 50 states, most of which had paper systems to route resources.
This year represented the first real test of the national system. When a long fire season ended last month, ROSS had allocated and managed more than 150,000 resources such as people, engines and machines across 50 states, says ROSS project manager Jon Skeels.
According to Skeels, ROSS relies heavily on the Java programming language to tie together a wide range of applicationsOracle databases, IBM Websphere and Hyperion's Brio business intelligence softwarefound in corporate systems.
Skeels, however, is hostage to the machinations of Congressional funding. When federal budget deliberations dragged on, Skeels had to postpone a migration to the 9i edition of Oracle's database system as well as plans to add storage area networking and processing power. Those upgrades will occur before the 2004 fire season.
ROSS users paid the price over the summer when the existing system crashed for 3.5 hours in June, from a database overload. Meanwhile, network performance was abysmally slow in June because there was "too much stress on too-small servers," says Skeels. That fact prompted many dispatchers to go back to the friendly confines of faxes and paper forms.
The culprit for the poor performance: an incorrect configuration setting in an Oracle database that keeps a running diary of requests. "There were a few days where ROSS was a four-letter word around here," says DeneenCone, dispatch center manager for the state of Arizona, which was dealing with a slew of wildfires as performance lagged. "We were using a dual system of ROSS and paper to be fail-safe, so it still worked out."
In Oregon, the slow performance led Randall Baley, dispatch coordinator for the Oregon Department of Forestry, to revert to "slow-time" from real-time. "We were just more comfortable with paper in the heat of battle," he says. "ROSS has potential, but the speed of the system wasn't enough.''
So how slow is ROSS? In theory, with ROSS, dispatchers take reports from the scene, check availability of resources, place orders for new equipment and personnel, and that's it.
By contrast, Baley's paper system involves filling out five different forms to request equipment, supplies and crews. These forms are then faxed or the contents read over the phone to a regional fire coordination center, where details are manually re-entered on forms. From there, details are again faxed or transcribed to a state coordination center.
Baley says the paper system can take at best a half hour to get an air tanker headed toward California. Requests for engines and supervisors could take "numerous hours" to fill. If running properly, ROSS can process that air tanker order in minutes because data is only entered once. The time frame for other resources depends on availability.
Skeels acknowledged the problems, which he contends have been corrected with additional hardware and a database fix. He also installed a plan of action to prevent future problems. "During that time [ROSS users]went back to paper," says Skeels. "No additional acres were lost."
Cone say ROSS functioned as promised after its June problems. Skeels is trying to quantify the returns from ROSS by tracking time and the number of orders moved.
Based on surveys with users, Skeels estimates ROSS led to better than a 50% decrease in the time needed to take and process orders. Skeels didn't have an average time for the paper system, but dispatchers like Baley say the time to fax something five times spans hours. Skeels says 2003 was largely a year to gather and create a baseline set of statistics for ROSS.
In the meantime, Skeels will continue to develop ROSS and fight a cultural war with champions of a slow, decades-old yet effective paper system.
"All ROSS is doing is taking what we were doing all along and automating it," says Cone. "Most people see it as all bells and whistles, but it really updates an archaic system. A lot of old fire dogs think it'll fail for a while, though."