Juice, a distributed task scheduling cloud system based on Mesos Framework developed by HUjiang Learning System Project Team (LMS), is officially open source today


Related links:


  • Juice: Evolution of Task scheduling system.

  • Juice open source address: https://github.com/HujiangTechnology/Juice


Why Juice

Juice is a set of distributed task scheduling cloud system based on Mesos Framework developed by The Learning System Project Team (LMS) of Hujiang. Based on this system, any job-type task scheduling can be realized.

Time-consuming computational task processing has been an ongoing requirement in Hujiang, such as multimedia transcoding, Map Reduce, intensive computing tasks, etc. Before Juice, different project teams developed different task processing systems, which cost not only people but also servers. Juice takes advantage of the idle computing power of the Mesos cluster and is responsible for unified interface return and task query functions, with the following advantages:

  • Based on the Mesos cluster, to maximize the use of idle resources to complete time-consuming calculations

  • Asynchronous scheduling, and provide results query and callback methods to obtain the results of execution

  • In distributed management, all time-consuming tasks are scheduled stateless in resource pools.

  • Task query: Currently, you can only query the status and result of tasks through interfaces.

The Juice open-source code is hosted on GitHub.