Computer Science 4611 : DBMS
Assignments 3-5 Overview

Introduction

During this semester we will be implementing several parts of an database management system. The system will basically have five components, and we will implement the middle three components. The five components of the system are:

1. Disk Space Manager
2. Buffer Page Manager (Assignment 3)
3. B+ Tree Files and the System Catalog (Assignment 4)
4. Query Processor (Assignment 5)
5. Query Language Parser

Each of these systems will make use of the earlier layers. We will have test routines for each layer and a final query system that will use the resulting entire system. Below I will give a short description of each layer.

Disk Space Manager

This layer will be provided to you. This layer will provide the basic routines for requesting, reading and writing pages that will make up files. We will be providing this layer to you largely because the code in this layer is very system specific and because we will be using a very simplified approach to maintaining files (our "files" will actually be parts of one large database file).

Buffer Page Manager

This layer will provide mechanisms to allow layers to request pages, deallocate pages, and pin (and unpin) pages in the Buffer layer. This provides an intermediate level between the file managing layers and the actual system layer that lets you abstract away details of page allocation (and when to read/write pages).

B+ Tree Files

In the first part of assignment 4 you will provide mechanisms for creating database files as B+ tree files. This layer will provide routines for creating and destroying a B+ tree file, adding records, deleting records, updating records, and opening a scan on a B+ file.

System Catalog

In the second part of assignment 4 you will implement a database catalog consisting of a relation relation (containing information about what relations exist in our system), an attribute relation (containing information about the attributes of that relation), and an index relation (containing information about hash indexes). You will include routines for setting up the initial catalog, adding and dropping relations from the system, adding and deleting tuples from a relation, showing the tuples of the relation.

Query Processor

In this layer we will add routines for creating iterators/scans for select, project and join queries (based on the information in the catalog).

Query Language Parser

The query language parser provided will allow us to issue queries in a simple query language that will then be executed by our query processor layer. This code will be provided. In a more complete system, a parser would be needed for the language we are using (likely a variant of SQL), we would then add a planner layer and an optimizer layer that would optimize the iterators/scans produced by our query parser.

Other Files

You will also find a few other files in the system not directly part of any of the layers shown above. These are:

• minidb.h and minidb.C - these define some basic classes for the code and deal with the debugging system
• system_defs.h and system_defs.C - the class SystemDefs represents all of the pieces that make up our database. A member of this class represents our whole database. We allocate this class at the start and delete it at the end of all of our testing programs.
• new_error.h and new_error.C - implements an error protocol system that allows the user to find out lots of information about what went wrong in the system. Check out the code provided to you for samples of how to use this code.