RACE

What is RACE?

framework to build event driven applications that

• can import/export from/to external systems - connectivity
• can process high event rate and data volume - scalability
• have extensible set of concurrent, low overhead components
• support distributed and massively concurrent operation

Actors - Basic RACE Components

• well known concurrency programming model since 1973 (Hewitt et al)
• Actors are objects that communicate only through async messages
  ⟹ no shared state
• objects process messages one-at-a-time ⟹ sequential code

Actor Systems - Configuration

• RACE actor systems are JSON configured graphs

  • nodes are actors
  • edges are pub/sub (bus) channels through which actors communicate

RACE - Implementation

• Master actor: initialization, supervision and termination of configured actors
• can model time
• local and global (network) bus

Example 1 - Live/Recorded NAS Visualization

• full NextGen SWIM (System Wide Information Management) import
• 500+ incoming network msg/sec (mostly JMS)
• real time ADS-B import from local antenna
• visualization with NASA WorldWind

Example 2 - SWIM Analysis

• used to analyze and verify SWIM data:

  • statistics (msg rate, volume, peaks)
  • XML validation
  • serve reports as HTML
  • detect track update anomalies (out-of-order,ambiguous,dropped,..)

Example 3 - LVC Simulation

• import live and (interactively) simulated tracks
• detect and send proximities to simulator
• detect NMACs of live tracks caused by simulator

TL;DR

• RACE is a actor based application framework
• primary design goals: scalability and extensibility
• can be used as a library (from external projects)
• used for runtime monitoring, trajectory analysis, simulation, web-serving dynamic data,...
• open sourced under Apache v2
• written in Scala, heavily based on Akka library
• ~300 files, ~21000 ncloc
• developed since 2015