TU Delft
 
Alexandru IOSUP
MSc projects
Parallel and Distributed Systems
EWI PDS A.IosupMSc projects
 
 
 
 
 
 
 
 
 
New! M.Sc. projects available!



About

This page contains information about doing a master's project supervised by Dr. Alexandru Iosup, as a part of either the Computer Science Master's Program, the Computer Engineering Master's Program, or the Embedded Systems Master's Program at TU Delft. Unless otherwise stated, you will graduate with a Master of Science title from the Parallel and Distributed Systems (PDS) group of the Faculty of Engineering, Mathematics and Computer Science (EWI), TU Delft.

The MSc coordinator of the PDS group is D.H.J. Epema. If you intend to do a master's project in the PDS group, you are advised to contact him at the start of your master's program to have your list of courses approved. In addition, you should contact him two to three months before you plan to start your research assignment or your master's project.

For more information on doing a master's project in the PDS group please consult the PDS group MSc projects page.




Bootstrapping our collaboration

I am always interested to discuss with students about the perspective of an MSc project within the PDS group, possibly not under my guidance. However, for our discussion to be meaningful, please read the following before contacting me or the MSc program coordinator, Dr. Dick Epema.

Your help is needed. If you are a prospective MSc student and would like to work with me you can contact me directly at any time, with or without a specific topic in mind-I usually have several MSc projects from which you can select, and the PDS group has many more. However, I can only supervise 2-3 MSc projects at a time, and the same probably applies for each of my colleagues; to deal with the number of requests I apply an equal-opportunity First-Come-First-Served policy.

I want you to enjoy the process of getting your MSc title (and have a personal life, too). To this end, I will be available with advice and encouragement, I will steer you gently towards getting your MSc title, and, perhaps most importantly, I will tune the process for your future aspirations. If you want to go to the industry, I will help you create a portfolio also reach results that are meaningful for your future employer. If you want to start a doctoral track, I will make sure you are equipped with the right tools at the end of your MSc project.

I am a results-oriented supervisor. This means that I don't mind if you make mistakes during the course of your MSc project; in fact, I encourage you to explore fearlessly and as much as possible during this period-it may be your last chance before reality hits back. Since the end result is what matters, I am open to discuss with you about a flexible work program. As a drawback, I will not settle with low-quality MSc thesis or poor work.

I do not mind helping you excel. While the first goal of an MSc project is to get an MSc title, I am going to try to help you achieve your full potential, should you desire it. If you are an ambitious and energetic student, I will help you keep the pace high.

For more information on doing a master's project in the PDS group please consult the PDS group MSc projects page.




Projects

My area of expertise is resource management in large-scale distributed computing systems. Thus, I will gladly coordinate MSc projects involving massively multiplayer online games (MMOGs), grid and cloud computing, peer-to-peer systems (P2P), performance evaluation at large scale, etc.

  1. KOALA-C: A Scheduler for Integrated Multi-Cluster and Multi-Cloud Environments
    Status: Completed by Lipu Fei, Sep 2013.
    Area: cloud scheduling, KOALA, IaaS clouds, scientific computing, performance evaluation, experimental work, real-world experimentation, simulation.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate a cloud-based scheduler of scientific workfloads, where tasks have unknown durations yet efficiency is a key property of the scheduling system.
    Summary:
    summary from the TU Delft Repository.
    Online MSc thesis:  KOALA-C: A Scheduler for Integrated Multi-Cluster and Multi-Cloud Environments.
    Achievements: adapted KOALA to use cloud resources, adaptation of two scheduling policies for efficient execution of tasks with unknown durations in multi-cluster and multi-cloud environments. Evaluated the performance of the proposed designs in both simulation and real-world experimentation. Good trade-offs and ability to improve performance tenfold vs traditional, unadapted scheduling policies.
     
  2. LudoGraph: a Sampling Capable Cloud-Based System for Large-Scale Graph Processing, Based on the Pregel programming model
    Status: Completed by Marcin Biczak, Jul 2013.
    Area: Pregel, MapReduce, graph processing, gaming analytics, big data, cloud-based data processing, sampling algorithms, performance evaluation, experimental work.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate a cloud-based, Pregel-based graph-processing platform, and to compare it with several well-known graph-processing platforms (Hadoop, neo4J, Stratosphere, GraphLab, etc.)
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  LudoGraph: a Sampling Capable Cloud-Based System for Large-Scale Graph Processing.
    Achievements: compared the performance of several large-scale graph-processing platforms, designed and implemented a cloud-based graph-processing platform based on the Pregel programming model, extensions to the Pregel programming model, experiments with sampling algorithms.
     
  3. A workload model for MapReduce
    Status: Completed by Thomas de Ruiter, Jun 2012.
    Area: MapReduce, workload characterization, workload modeling, scheduling, performance evaluation, performance modeling, trace-based simulation.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate a workload model for MapReduce, starting from several real-world workloadtraces.
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  A workload model for MapReduce.
    Achievements: several models verified against four real-world MapReduce traces collected in some of the largest MapReduce users in the world (e.g., Google).
     
  4. Tribler-G: A Decentralized Social Network for Playing Chess Online
    Status: Completed by Egbert Bouman, Mar 2012.
    Area: Tribler-G, online gaming, peer-to-peer systems, distributed systems, protocol development, performance evaluation, empirical, user experience.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate Tribler-G, a peer-to-peer social network for playing chess and other board games online.
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  Tribler-G A Decentralized Social Network for Playing Chess Online.
    Achievements: Tribler-G web site Tribler-G is fully functional.
     
  5. Performance Evaluation of Cloud Infrastructure using Complex Workloads
    Status: Completed by Athanasios Antoniou, Jan 2012.
    Area: SkyMark, testing, complex workloads, cloud scheduling, provisioning, allocation, heuristics, performance evaluation, empirical.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate SkyMark, a system for assessing the performance of IaaS clouds using complex workloads.
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  Performance Evaluation of Cloud Infrastructure using Complex Workloads.
    Achievements: study of the overheads that the cloud software stack imposes to the workload execution; analyze the performance and cost of six provisioning and three allocation policies through experimentation in three IaaS environments, including Amazon EC2.
     
  6. WebHack: a research system for social massive multiplayer online games
    Status: Completed by Arnoud Bakker, Jan 2012.
    Area: online gaming, NetHack, distributed system, scheduling, massively multiplayer online gaming, online social gaming, performance evaluation, empirical.
    One-line description: the goal of this M.Sc. project is to design, implement, and evaluate a Facebook-integrated online gaming system that can support massive amounts of players.
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  WebHack: a research system for social massive multiplayer online games.
    Achievements: supporting 300,000 NetHack players through the Facebook platform.
     
  7. Trace-based Performance Analysis of Scheduling Bags of Tasks in Grids
    Status: Completed by Shanny Anoep, Dec 2008.
    Area: grids, scheduling, bags-of-tasks, the Grid Workloads Archive, performance evaluation, trace-based simulations.
    One-line description: the goal of this M.Sc. project is to design mechanisms and tools for analyzing the performance of grid scheduling algorithms.
    Summary: summary from the TU Delft Repository.
    Online MSc thesis:  Trace-based Performance Analysis of Scheduling Bags of Tasks in Grids.
    Achievements: co-author for top conference paper, journal paper, tools for the Grid Workloads Archive.
     

 

   Notes
---
---

     

Last modified: Sun, 4 November, 2012 1:38 AM
The newest version of this page can be found at: http://www.pds.ewi.tudelft.nl/~iosup/msc_projects.html
Copyright © 1998-2009 Alexandru Iosup. All Rights Reserved.
Google Analytics .