This project is an educational job scheduler based on the work done by Y. C. LEE @Macquarie University in the COMP3100 Distributed Systems Course. The ds-sim code is included as a submodule and found here.

The stage1 and stage2 tests are also included as a submodule, hosted in a separate repository.

• Get the dependencies:

# git submodule init --recursive

• Build the scheduler, reference implementation of ds-client and the dependency ds-server:

# make all

• Run a test case against the scheduler:

# make test

• Run the university-provided test-suite for stage 2:

# make run-demo

Design and implement at least one new scheduling algorithm that optimises average turnaround time.

In particular, your scheduling algorithm as part of your client shall schedule jobs to servers aiming to minimise the average turnaround time without sacrificing too much of other performance metrics, i.e., resource utilisation and server rental cost.

As these objectives are incompatible (conflicting performance objectives), the optimisation of one objective might lead to sacrificing the optimisation of other metrics. For instance, if you attempt to minimise average turnaround time by minimising waiting time, you may end up using more resources resulting in high costs (i.e., server rental costs).

Therefore, you have to define the scheduling problem clearly indicating your performance objective(s) and justifying your choice. In addition, you must discuss your algorithm and scheduling results in comparison with four baseline algorithms (FF, BF, WF and FC) and their results.