Powering Oceans 3.0: the Low Latency Computing Environment Developed by Ocean Networks Canada

Ocean Networks Canada (ONC) operates a comprehensive suite of ocean observatories that continuously stream vast quantities of oceanographic, physical, and biochemical data. A critical component of this system is the need for low-latency computing, particularly when using MATLAB for processing. Traditional approaches proved inadequate for the required low-latency applications. To address this, ONC introduced “MATLAB-as-a-service,” an in-house solution to eliminate MATLAB startup latency by maintaining a pool of running MATLAB instances. MATLAB-as-a-service integrates with ONC's Oceans 3.0, significantly enhancing the responsiveness of data processing tasks, such as automated testing, which is now completed in three hours instead of eight. This paper outlines the key architecture, components, and operational benefits of MATLAB-as-a-service, demonstrating its effectiveness in overcoming latency challenges when using MATLAB for processing.