Description
In our research lab, we are researching novel stream-processing geo-distributed systems, that can be used to implement and execute modern multi-modal applications. In the last years we have developed an advanced and easy to program framework called World Wide Streams. In a bid to further improve fault-tolerance and, more in general, lifecycle management of geo-distributed applications, we want to implement a state-of-the-art heartbeat-based solution on our current software stack to monitor the state of workers and the applications running on top of them. The core idea would be to make the system always aware of the state of the different distributed operators and their workers, enabling re-execution and other fault-tolerant strategies.
Duration: flexible, to be agreed. Can be combined with a Master Thesis.
Responsibilities
* Familiarize with the World Wide Streams platform and its current lifecycle management
* Read state-of-the-art and explore how other platforms (e.g. Flink) design their lifecycle management using heartbeats
* Extend worker components to emit heartbeats on an own execution thread
* Extend the lifecycle management components to use heartbeats for decision making
Qualifications
* Student enrolled in a Master of Computer Science/Engineering in a Belgian university
* Strong programming skills in Typescript and understanding of Python
* Good knowledge of distributed systems, preferably streaming platforms
* Familiarity with open-source software and tooling, including CI/CD pipelines
This is an unpaid internship for Master students.