masterDB

Implementation of a buffer pool based database system. The development of the system was carried out over the duration of one year as part of a university project at the Chair of Applied Computer Science III, University of Mannheim.

Table of Contents

• Prerequisites
  • System Requirements
• Project Structure
  • src/buffer
  • src/cli
  • src/infra
  • src/interpreter
  • src/main
  • src/partition
  • src/segment
  • src/threading
• Getting Started
  • Quick Start Guide
• Project Status

Prerequisites

There are several tools you will need to install and execute the application. In the following is a list with all required tools and technologies required for installing and running the system:

• UNIX-like operating system: All development and testing was carried out on UNIX-like operating systems. As various UNIX system calls are used throughout the program, the system will not run under Windows.
• GNU Make: As development and testing was carried out on UNIX-like operating systems, the de-facto standard build and install system GNU Make is used.
• C++17 compatible compiler: As C++17 features are frequently used and all tests were carried out with GCC 7.3.0, we recommend to use this compiler or a never version
• Intel Processor: As we use various Intel intrinsics, the system is currently only able to run on Intel processors
• Little Endian Architecture

Project Structure

buffer

Implementation of the buffer pool.
For further informations, take a look at src/buffer/README.md

cli

Implementation of a command line interface. By executing queries via the command line interface, the system can be controlled.
For further informations, take a look at src/cli/README.md

infra

Various helper files and functionalities.
For further informations, take a look at src/infra/README.md

interpreter

Implementation of various low level page interpreter working on the bits and bytes of pages.
For further informations, take a look at src/interpreter/README.md

main

Implementation of the database instance manager and main-method.
For further informations, take a look at src/main/README.md

partition

Implementation of raw partitions (partitions for raw/block devices), file partitions (working on files in the file system) and the partition manager (managing all partitions).
For further informations, take a look at src/partition/README.md

segment

Implementation of free space management segments (segments with a free space management of pages) and segments for the N-ary storage model (NSM).
For further informations, take a look at src/segment/README.md

threading

Implementation of the threadpool.
For further informations, take a look at src/threading/README.md

Getting Started

The build and installation process will be described in the following. Follow the Quick Start Guide for a fast installation and get the system running.

Quick Start Guide

• Make sure all the Requirements are satisfied

• Clone the source with git :

git clone https://github.com/WeberNick/masterDB.git

• Build and install:

cd masterDB/src
make

• Run

./main/main

• Getting Help

./main/main --help

Project Status

Already implemented

• Partition Manager: Maintains all the different partitions of the database system
  • File Partitions: Using files managed by the filesystem
  • Raw Partitions: Using a raw device without any filesystem on it
• Segment Manager: Maintains all the different segments in the database system
  • Free Space Management (FSM) Segments: Segments with a free space management of their pages
  • Slotted Pages (SP) Segments: Segments with slotted pages
• Buffer Manager: Maintains the buffer pool of the system. On memory pressure, a random eviction policy is applied
• Threading: The system is multi threaded using a thread pool
• Interpreter: Various interpreter are implemented to work with different page layouts
• Tracing: Tracing can be activated by setting a command line argument
• Command Line Interpreter: The system is controlled via a command line interface

Todo

• Implement higher level functionalities such as a query execution engine
• Implement a lock manager and transaction manager
• Implement index structures
• ...