Project Ideas
Please read all of the following documents to get ideas for viable projects. I would encourage all of you to choose one of the following projects. I will allow other project ideas, but you have to discuss it with me before writing it up (to make sure both relevance and difficulty is appropriate). I encourage groups of 2 for these projects, although some projects might work with 1 person, or some might require 3 poeple because of the ambitious nature of the project. In certain cases, I will allow more than 3 students in a group, but it highly depends on the project and nature of the work.
| Fabriq:Bench | Benchmarking mainstream Distributed Message Queues |
| Fabriq:Sched | Leveraging Fabriq as a building block for Distributed Scheduling with CloudKon+ |
| FusionFS:CKPT | Efficient Checkpointing with Distributed File Systems |
| FusionFS:Hadoop | Improving Hadoop through FusionFS |
| FusionFS:Lib | Improving FusionFS Performance through User-level Library Interfaces |
| FusionFS:Sched | Data-Aware Scheduling for Distributed File Systems |
| GeMTC:Hadoop | Supporting Hadoop Applications on Accelerators |
| GeMTC:MIC | Supporting MTC Applications on Intel Xeon Phi Many-Core Accelerators |
| GeMTC:Mon | Monitoring GeMTC on NVIDIA Hardware |
| MATRIX:Bench | Benchmarking the state-of-the-art Task Execution Frameworks of Many-Task Computing |
| MATRIX:Hadoop | Improving Hadoop Performance and Scalability through Distributed Data-Aware Scheduling |
| MATRIX:HadoopSim | Understanding the Scalability of the Hadoop Framework through Simulations |
| MATRIX:Swift | Accelerating Swift Many-Task Computing Applications through Distributed Data-Aware Scheduling |
| NET:MPNet | Improving Network Throughput through Multipath Network Routing Systems |
| NET:MPSim | Improving Network Throughput through Multipath Network Routing Simulations |
| OS:DistOS | Light-weight Distributed Operating Systems |
| OS:Power | Understanding power management and its impact in modern Intel Haswell Processors |
| Slurm:Bench | Benchmarking of the state-of-the-art Job Management.Systems for High Performance Computing |
| Slurm:Data | Exploring Data-Aware HPC Scheduling with Slurm |
| ZHT:Bench | Benchmarking mainstream NoSQL databases |
| ZHT:Graph | Design and implement a graph database on top of ZHT |
| ZHT:Mon | Investigation of the Communication Overheads for Distributed Monitoring with Aggregation Tree |
| ZHT:YCSB | Yahoo Cloud Serving Benchmark on ZHT |
CS550