For your final project you will investigate a research challenge in distributed systems. We will allow two types of projects:

Research Focused: Your group must pick a problem under active research, read papers in the area, and implement a solution to the problem. Your solution can be a re-implementation of an existing research project or a combination/extension of current approaches. You will need to build and evaluate the performance of your system to compare it against prior research. If your focus is more algorithmic, you may want to build a simulator instead of a real platform so you have better control over the environment.

Implementation Focused: Your group must build a simplified version of an existing distributed system. You still will need to read research papers about the system you are implementing, but your effort will be focused more on implementing the system and less on evaluating and comparing it against other proposed approaches.

You have about two months to complete the project so it should be more significant than a standard programming assignment, while still being feasible to complete within that time frame. The project will be broken into milestones.

Due Dates:

Milestone 0: Form a team

  • Due 10/12 - earlier if possible

You must form a team of 3-4 students. We suggest teams have 4 members if possible. All members are expected to contribute equally to the project, although different students may have specialized roles. However, all students should at least partially contribute to both the coding and writing portions of the project. One student should be selected as the team leader who will act as the main contact point of the group.

When you form a team you do not yet need to have selected a project idea, however it is best if you decide an area (or areas) you want to work in and be sure all team members agree. See the sample projects under Milestone 1 for more details.

Submission Instructions:

  • First, pick a team name. Your team name must be that of a famous scientist, e.g., Galileo or MarieCurie. Pick someone interesting, inspiring, and unique.
  • The team leader should create the team repository using GitHub Classrooms, entering the chosen team name. Make sure that nobody else has already picked your name! Other team members should then join the team. It is hard for us to delete teams or move people between them, so be careful and don’t join the wrong team! We may deduct points if you cause us extra work.
  • Fill in the file in the repository with your team members, team name, and optionally a brief description of the project idea(s) you are considering
  • Create an Issue in your repository once your readme is complete. The title of the issue should be Milestone 0 - TEAM NAME (with your team name filled in). The body of the issue should list the team member names with leader first.

Milestone 1: Select a topic

  • Due 10/19 - earlier if possible

We strongly encourage you to complete Milestones 0 and 1 early. If you submit Milestone 1 at least 4 days early you will receive a 5 point bonus to your report grade.

For this milestone you will need to submit a 1-2 page report that answers the following questions:

  • What is the topic area of the project? Why is it important and what specifically will you build or design?
  • What are the main distributed systems challenges you will face? (see our slides for the list of distributed systems challenges)
  • What are some sample papers (at least 4) that will guide your work?

All of your milestone reports will be judged partially on how well they are written, so write complete sentences/paragraphs, not just a bulleted list!

We will not permit two groups to work on the same project idea. Precedence will be given to whichever group submits their proposal first. Your project proposal for Milestone 1 must be approved by the professors before you can submit the following assignments! If you submit early we will try to approve your project early.

Submission Instructions:

  • Your writeup must be in Markdown format and saved in the file milestone-1/
  • Create an issue with title Milestone 1 - TEAM NAME and the body should have a link to your report file.
  • Click the button on the top right side of the Issue view to Assign the issue to both professors

Sample Project Ideas - Research Focused

Linux CPU Scheduling: Modify the Linux CPU scheduler and evaluate how your changes affect the performance of different types of applications. For example, you could reproduce the code from Fewer Cores, More Hertz (Usenix ATC 20) which proposed techniques for adjusting the scheduler to perform better when the CPU performs dynamic frequency scaling.

  • Similar ideas: VM CPU scheduling on a hypervisor like Xen; Dynamic resource allocation for containers using cgroups; etc

Load Balancing: Evaluate the performance of different load balancing algorithms under a variety of workloads and server conditions. To evaluate load balancing approaches at large scale, you could build a simulator to test the performance of different algorithms. You should focus on a particular challenge, for example reproducing the load balancing algorithm in Prof. Wood’s NetKV (ICAC 16) project which dealt with skewed workloads arriving at a memcached cluster.

  • Similar ideas: Task scheduling in Map Reduce; website load balancing; etc

Optimization/Metaheuristics: Simulate and compare some metaheuristic or ML algorithms for Scheduling or placement considering several features such as workload, deadlines, priority, CPU/Memory cost, etc For example comparing MOPSO, NSGAIII, Neural Networks, Clustering or Classification based approaches.

Consensus: Consensus algorithms allow multiple nodes in a distributed system to reach agreement about a decision even if some nodes can fail. For example, Raft (Usenix ATC 14) provides an algorithm for ordering requests when nodes can crash or become disconnected. You could either implement Raft from scratch, or extend the official implementation in some way.

  • Similar ideas: Byzantine Fault Tolerance to handle arbitrary failures and malicious nodes

Container Migration: Virtual machine migration is a popular tool for managing data centers, yet container migration has not reached the same level of maturity. You could explore container migration approaches based on CRIU such as Voyager (ICDCS 17) and evaluate how they perform in different environments.

  • Similar ideas: Optimizing container or VM migration by eliminating redundant data transfers; resource management algorithms to decide when to migrate or which host to migrate to

P2P Simulation: Implement a simulator that mimics the behavior of a large P2P system such as bittorrent, gnutella, or bitcoin. Evaluate different algorithms for finding content or sending data in the network.

Sample Project Ideas - Implementation Focused

These project ideas focus on implementing a simplified version of a distributed system “from scratch”. Designing and building a full system like this is a lot of work, so you will need to limit your scope to a specific set of distributed systems problems.

Distributed Database from Scratch: Implement a basic distributed database. You should not spend a lot of effort on the database query engine (perhaps you can just reuse something simple like SQLite) and instead should focus on challenges such as supporting consistency and fault tolerance.

  • Similar ideas: Build a Distributed Hash Table like Chord or implement a simulation of a large scale P2P file sharing system.

Serverless Platform from Scratch: Implement a simplified Serverless / Function as a Service platform. A serverless platform dynamically starts and stops containers or VMs as requests arrive. The basic components include a gateway load balancer, a queueing system that can buffer requests, and an autoscaler that decides when to add or remove nodes.

RPC from Scratch: Implement your own RPC library. Your version of RPC could explore interesting distributed systems challenges like fault tolerance or scalability. For example, your RPC library could automatically load balance calls across several servers, or detect a server failure and reissue a request to a backup replica.

Bad Project Ideas

Projects like the following will not be approved:

  • Building a web application that combines several cloud services (e.g., a meeting room reservation system). This is a distributed system, but we want you to focus on the underlying concepts, not an application.
  • Taking code from an existing research project and measuring its performance. If code exists, you will be expected to modify / extend it in some way. If no code is available, then it is acceptable to try to directly mimic an existing project.

Milestone 2: Literature Review

“A literature review surveys books, scholarly articles, and other sources relevant to a particular issue, area of research, or theory, and by so doing, provides a description, summary, and critical evaluation of these works in relation to the research problem being investigated. Literature reviews are designed to provide an overview of sources you have explored while researching a particular topic and to demonstrate to your readers how your research fits within a larger field of study.”

– Fink, Arlene. Conducting Research Literature Reviews: From the Internet to Paper. Fourth edition. Thousand Oaks, CA: SAGE, 2014. See this article for more information on writing a literature review.

For Milestone 2 you must perform a Literature Review which summarizes and synthesizes information from key sources related to your project. You will present the importance of the project area, motivate the difficulty of the key challenges, and describe the state of the art in resolving them.

Article Selection: You must present information from at least 6 articles for 3 person teams or at least 8 articles for 4 person teams.

  • Use a tool like Google Scholar to find articles related to your topic area. Note that you may need to be on the GW VPN in order to access research article PDFs.
  • In general, we prefer you to use research articles such as those from conferences or journals published by ACM, IEEE, or Usenix, since they have more depth and are peer-reviewed (and thus more likely to be accurate) than random webpages.
  • At most half of your articles (preferably less) may be technical “blogs” or other articles found online. However, these should be thorough, in-depth articles, not just how-to documents or news articles designed for everyday readers. For example, this article on game network coding and performance is a good choice, but this article on cloud gaming and 5G would be a bad choice since it lacks technical depth. If you think someone without a CS background could easily understand the article, then it is probably a bad choice since it will not have sufficient technical detail to teach you anything.
  • Your articles should be recent. It is fine to have some “fundamental” papers from prior decades, but be sure to also include some recent work that accounts for the state of the art in the area.
  • You may find it useful to look at ACM Computing Surveys, a journal that publishes literature reviews in many CS areas. You might find one related to your topic and it will give you an idea of what an in-depth literature review looks like (yours will be much shorter).

Report Structure: Below is a suggested structure for your paper. You may use something different but must cover all these points.

  • Introduction: Give the background for your proposed project and the importance of resolving it. Refer to your articles to provide motivational data or use cases that show why this is something useful and important. (~0.75 page)
  • Challenges: Describe the key distributed systems challenges related to your project. Use the terminology defined in our class (scalability, heterogeneity, etc) and show how they can impact your project. Focus on the problems presented by the related work or any challenges which you feel the articles fail to address sufficiently. (~1 page)
  • Approaches: Summarize the designs and algorithms proposed by related work. Try to categorize them so that you group different types of approaches together in a way that adds meaning. (~1.5 pages)
  • Project Proposal: Very briefly outline how your project relates to the problems and solutions described above. For example, you could explain how you are focusing on a specific sub-problem or extending an existing piece of work. This section can be short: 1-2 paragraphs at most.

Format and Submission: Your report should be 2-4 pages long and follow these guidelines.

  • As with all writing, it is never acceptable to copy/paste from another author. All of your writing must be your own! If we find any sentences copy/pasted from another source (unless it is a direct quote), we may deduct 10-50 points.
  • Your document must be submitted as PDF and should be in the form of a technical article. You may want to use a document template, such as those provided by ACM, in Word or Latex format.
    • We suggest single column, justified text, size 10 font, 1.25x line spacing, or similar.
    • If you use Latex (not required), we encourage you to use a service like Overleaf to simplify collaboration and generation of the document.
  • Your writing must include citations where appropriate using a style such as [Wood18] to cite a paper by Wood et al in 2018.
  • Save your document as milestone-2/literature.pdf and create an issue with a link to the document assigned to the professors when you are ready to submit. While not required, we suggest storing copies of all the related articles you surveyed into your repository.
  • In consultation with the group members, the leader should prepare a Phase Time Sheet and save it as milestone-2/phase-2-time-sheet.xlsx
  • If any team members have concerns about how work is being divided up in the group or are experiencing problems with their partners, complete the Partner Feedback form

The output of this report will be the fundamental basis of the next Milestone. For your next Design report, you need to make a decision about the solution you will develop for your problem according to the pros and cons of the studies cited in your literature review. Your solution can be chosen among the cited studies or it can be a combination of them, or you might propose your own solution and method to address the issues in your problem space. But the important point is that your proposed method should be based on an understanding of what others have done in the field. For the Literature Review Milestone you only need to give a basic outline of your plan, which you will then fully develop in Milestone 3.

Milestone 3: Design Document

The deadline has been extended until 11:59PM Sunday November 8th.

The design document should be composed of sections as follows:


Start with a very brief introduction to the problem you are trying to solve, followed by a high level overview of your proposed project. You may be able to partially reuse some text from your Milestone 2 Introduction and Project Proposal sections for this purpose, but this section should be only 1-2 paragraphs long.

Project Design

This section should describe the key components and algorithms that will make up your project. Describe your architecture and how the different components will interact. Where possible, you should use the terms and design patterns that we have discussed in class, such as in Lecture 4 - Distributed Architectures. You also should specifically mention how your design allows the system to handle some of the core distributed systems challenges: Heterogeneity, Openness, Security, Failure Handling, Concurrency, Quality of Service, Scalability, and Transparency.

When describing your components and how they communicate with each other, you must include several diagrams as described below to illustrate your design. Not including figures, this section should be 2-4 pages long. You do not need to go into too much detail about what each component will do, but you give enough information that we will be confident you know how to proceed with your implementation.

Design Diagrams

There are many ways to show and explain the structure of your system. One of the best tools, to trace your architecture and algorithms, is UML diagrams. These diagrams enable the designers to explain their mindset and thoughts, to the developers, and also the QAs can execute test scenarios and test cases to find the possible flaws.

Since we would like to follow a standard strategy to produce a distributed system, we are going to use UML diagrams to illustrate the system’s components and interactions. One of your diagrams must be a Use Case style diagram to show the overview of your system functionality and different functions in your system. According to the nature and characteristics of the project, you are free to choose two or more additional diagrams among all UML types. For example, your Use Case diagram might show the overall system, then you can use the Deployment and Component diagrams to show the software and hardware components of your system. A Communication diagram also can illustrate the connections between the components, Sequence diagram explains the sequence of actions, events, and processes between the components, objects, and actors in your system. You can also use Activity and State diagrams to explain more detail about a process or object in your system. We expect to see at least 2 or 3 diagrams in level 0 and 1 to illustrate a proper overview of your system.

We have made you two examples to illustrate a small part of a hypothetical system to understand how can we use these diagrams to design our system. Moreover, we had an example of a sequence diagram for a small multi-threading system in the slides of the Thread-Process lecture.

Also please check the following articles to have a better understanding of UML diagrams and also modeling distributed systems with UML.
Please notice that you might see different annotations of UML in different resources but we strongly encourage you to check the annotations, definitions, and examples.

Project Schedule

The final section of the report must include your plan for completing the project by the end of the semester. We suggest breaking down your project into a series of prototypes, from the simplest possible demo up to your complete system. Each new prototype should add more functionality or complexity to your system. Finally, you should include some kind of evaluation plan describing how you intend to test the system’s correctness and performance. Be realistic; there is not too much time and we will expect you to complete what you propose on schedule.

You must include a timeline for when you will complete each of your prototypes, ideally in the form of a Gantt chart or similar. The work breakdown should also specify which team member will be responsible for each component or feature.

Format and Submission

Your report should be approximately 3-6 pages long. You should follow the same formatting template instructions given for the Milestone 2 report.

  • Save your document as milestone-3/design.pdf, then create an issue with a link to the document and assign it to the professors when you are ready to submit.
  • We suggest storing copies of your design diagrams and the source files used to generate the PDF (i.e., the word or latex files) in the repository too
  • In consultation with the group members, the leader should prepare a Phase Time Sheet and save it as milestone-2/phase-2-time-sheet.xlsx
  • If any team members have concerns about how work is being divided up in the group or are experiencing problems with their partners, complete the Partner Feedback form

Milestone 4: Final Report

Your final project submission must include a ~15 minute video and a final report.

Final Report

Your final report will primarily be a combination of the reports you wrote previously. It should be a “stand alone” piece of writing, meaning that a reader should be able to understand your full project without needing to look at the other documents. You should use the following layout which re-uses your prior reports and adds two new sections:

  1. Introduction: Background on the problem and your proposed solution (Reports 1, 2, and 3)
  2. Related Work: Descriptions of key related systems or papers (Report 2)
  3. Project Design: Explaining and illustrate your overall project structure and design decisions (Report 3)
  4. Distributed Systems Challenges (partially new section): This should be an extended version of what you wrote in Reports 1 and 3. You must list all of our key Distributed Systems design challenges (Heterogeneity, Openness, Security, Failure Handling, Concurrency, Quality of Service, Scalability, and Transparency) and discuss how they relate to your project. Even if you have not solved one of the challenges, you should still explain how it relates to your project and what would need to be solved for a real system deployment.
  5. Project Outcomes and Reflection (new section): In this section you should describe the overall result of your project and reflect on your design decisions. You should include some sort of measurements to understand how it performs or works (ideally in comparison to another system). For example, you could measure the performance of your system in terms of throughput and latency. Finally you should discuss how well your design choices worked and what you think the potential of your ideas are. You should be honest about your results – we don’t really care how well your approach does, but we do want to see that you have thought about its potential benefits and drawbacks.

Final Video

You should create a ~15 minute long video explaining your project and demoing how it works. The goal of your video is to “sell” your project to us and get us excited about what you have done. Your presentation should be understandable to anyone who has taken our distributed systems course.

We suggest the following structure:

  1. Introduce your project and the problem that it solves.
  2. Briefly present related work/systems.
  3. Describe the key design choices of your system, with a discussion of how they relate to some of our distributed systems design challenges.
  4. Show us some (NOT all) of your code and explain how the code relates to your design. For example, pick one or two key components and show us the main functions, classes, or data structures.
  5. Demo a portion of your project running. You do not need to show every part of your project, just a key aspect. For example, if your project focuses on reliability, show what happens when you cause a crash.
  6. Discuss the overall impact of your design and present some evaluation of how well it works.
  7. Conclude your presentation with a summary of your project’s key points.

Note that this is a lot of content to cover in 15 minutes, so you will need to be brief and only focus on the main points. If your video is more than 20 minutes we may deduct points.

The simplest way to create your video is to have your team members join a Zoom/Skype call and record as your present. While we expect your presentation to look good (i.e., using clearly readable, nicely structured slides), we do not expect you to do any fancy video editing.

Submission and Deadline

Your final submission must be made by December 22nd, 5PM. However, we will award 1 extra bonus point to your project for each day early that you submit, up to 8 points extra if you submit by the original deadline of December 14th 11:59PM.

To submit, complete the following steps:

  • All of your code and written reports should be in your GitHub Repository. The final report should be in the milestone-4/ folder.
  • Your video should be uploaded using this form. You should not include your video in your github repository since it is a large file.
  • In consultation with the group members, the leader should prepare a final Phase Time Sheet and save it as milestone-4/phase-4-time-sheet.xlsx
  • If any team members have concerns about how work is being divided up in the group or are experiencing problems with their partners, complete the Partner Feedback form. Ideally you should do this as early as possible!
  • When the above steps are complete, create an issue in your repository and Assign it to both of the instructors to notify them that you are ready for grading. Include a link to your final report in your issue.

After that, have a great break and a successful future building distributed systems! We hope you have enjoyed the semester and learned a lot!