Skip to content

Latest commit

 

History

History
137 lines (103 loc) · 6.2 KB

README.md

File metadata and controls

137 lines (103 loc) · 6.2 KB
Raw

MRBox 🎁

A user-friendly application for the management, processing and analysis of Big Data across local and remote platforms.

Table of Contents

  1. About the Project
  2. Getting Started
  3. Usage
  4. Documentation
  5. Contributing
  6. License
  7. Acknowledgements

About the Project

The analysis and processing of big data often requires combining remote and local platforms, services and cloud infrastructures. Due to the heterogeneity of these resources and the cost of transferring large-scale data, performing data-intensive tasks across them is typically a non-automated and laborious undertaking or requires the knowledge of specialized workflow tools by non-experts.

The purpose of this application is to provide a high-level and general picture of the local and remote infrastructures, as well as to enable the user to schedule, check locally, execute and remotely track operations on data residing at any integrated platform.

The application is structured according to the filesystem hierarchy of the operating system so that the monitoring will take place through seemingly local files and folders. The initial goal is to integrate Hadoop and B2DROP services and to be able to schedule MapReduce jobs with possibilities to expand to other platforms and types of data operations in the future.

High Level Design

Features

The main features currently implemented are:

  • Monitoring of the local /mrbox folder to track the creation, modification, deletion and relocation of its files and sub folders

  • Integration of Hadoop and implementation of one way synchronization between the local /mrbox folder and the /mrbox folder of the remote HDFS

  • Simple scheduling of Map Reduce jobs from the local file system to be executed in a remote HDFS

  • Automated fetching of Map Reduce jobs' outputs locally, after their completion remotely

  • Implementation of the .link file type. When the output generated by a Map Reduce job is larger than the file size that can be stored locally, a .link file is created locally with a link to the remote HDFS. The link can be used in the browser to view the output in HDFS

  • Integration of the B2DROP platform using the ownCloud Desktop Client.

Built With

Getting Started

Prerequisites

To use the app with a remote HDFS you need:
  1. A remote Hadoop cluster that you can connect
  2. The same version of Hadoop installed in your local machine
  3. To update the src/core/mrbox.conf file with:
    • the inet of the remote Hadoop cluster (hdfsHost)
    • the port of HDFS (hdfsPort) in the remote cluster
    • the path of the Hadoop installation locally (hadoopPath)
    • the local path to the /mrbox folder (localPath)

Additionally, in the mrbox.conf you can change the path where the /mrbox folder will be created on HDFS (hdfsPath) and the size limit of the files that will be fetched locally from HDFS or other remote sources (localFileSizeMB).

To synchronize your local /mrbox folder with a folder in B2DROP:

Follow the instructions here to create a B2ACCESS account and mount B2DROP using the WebDAV client. Prior to run the MRBox application, start the ownCloud Desktop client on your local machine.

Installation

To install Miniconda:
make setup
To create a conda environment and install the project's requirements:
make build

Usage

To run the application:
make run
To schedule a Map Reduce job to be executed in the remote Hadoop cluster:
  1. Write a Mapper and a Reducer to your preferred programming language
  2. Create a .yaml file outside the local /mrbox folder that contains:
    • mapper: absolute path to the Mapper's file in the local file system
    • reducer: absolute path to the Reducer's file in the local file system
    • input: path of the input data, relevant to the local /mrbox folder
    • output: path where the outputs will be created, relevant to the local /mrbox folder A template for the .yaml file can be found here.
  3. Move the .yaml inside the local /mrbox folder.

The Map Reduce job will be issued to the remote cluster. After its completion, the outputs will be automatically created in the local output path specified. In case of large output files, a .link file containing their HDFS path will only be created locally. You can use this link in your browser to view these large output files on HDFS.

[to add about file commands on files]
For help:
make help

Contributing

In open source, we feel strongly that to really do something well, you have to get a lot of people involved. ~ Linus Torvalds

Any contribution is highly valued and appreciated. To contribute:

  1. Fork the project
  2. Create your feature branch: git checkout -b feature/<newFeatureName>
  3. Commit your changes: git commit -m 'Add <newFeatureName>'
  4. Push to the branch: git push origin feature/<newFeatureName>
  5. Open a pull request

License

Distributed under the Apache, Version 2.0 License. See LICENSE for more information.

Acknowledgements

This project was developed with the support of Mr. Iraklis-Angelos Klampanos at the National Centre for Scientific Research “Demokritos” during the intership program of the National Technical University of Athens.