ECS 165B Spring 2011 - Database System Implementation

DavisDB Project

1. Record File Management Component. In this part, you will implement a set of functions for managing unordered files of database records. This component will rely on a lower-level Paged File component that we will provide. The Paged File component provides an in-memory buffer pool and performs low-level file I/O at the granularity of pages.

2. Indexing Component. In this part, you will implement support for building B+-tree indices on records stored in unordered files.

3. System Management Componenent. In this part, you will implement a potpouri of database and system utilities, including data definition commands and system catalog management. This component will rely on the Record Management and Indexing components, as well as a command-line parser which we will provide.

4. Query Engine Component. In this part, you will implement a query evaluation engine for a fragment of SQL involving select-project-join queries and updates. This component relies on the three earlier components, and will also use the provided command-line parser.

Build platform. The standard build platform for DavisDB will be the 32-bit Linux workstations in the CSIF labs in the basement of Kemper Hall. You are free to develop your code on whatever platform you wish (e.g., Mac OS X laptop, Windows machine running Cygwin), but problems you encounter on such platforms will not be supported, and it is up to you to ensure that your implementation builds and runs properly on the lab machines.

Debugging. Use of gdb or an IDE with integrated debugger, such as Eclipse with CDT, is encouraged for this class. There are currently some issues with the version of Eclipse installed on the lab machines; see here for a temporary workaround that should get Eclipse working.

Subversion. We also will require all students to coordinate their team efforts, and submit their project components, via subversion. Subversion repositories have been set up for each group on the CSIF machines at /home/cs165b/CSIF-Proj/cs165b-X/trunk/DavisDB, where X is the number of your group. Access to the repositories is restricted (via Linux permissions) to the members of your group, the instructor, and the TA. To get started, from the CSIF machines, type svn co file:///home/cs165b/CSIF-Proj/cs165b-X/trunk/DavisDB (replacing X with the number of your group). For example, user green in group 0 would type this and see:

  [green@pc12 ~]$ svn co file:///home/cs165b/CSIF-Proj/cs165b-0/trunk/DavisDB
  A    DavisDB/RecordFileHandle.h
  A    DavisDB/PageFileHandle.h
  A    DavisDB/PageFileManager.cpp
  A    DavisDB/RecordManager.cpp
  ...
  A    DavisDB/Common.h
  Checked out revision 3.

  laptop:~ tjgreen$ svn co svn+ssh://green@pc12.cs.ucdavis.edu/home/cs165b/CSIF-Proj/cs165b-0/trunk/DavisDB
  A    DavisDB/RecordFileHandle.h
  A    DavisDB/PageFileHandle.h
  ...
  laptop:~

If you have already started coding the first project component, after checking out the code from the repository, you will want to copy your versions of any changed files into the created directory, commit your changes via svn commit, and work in that directory from now on. Some basic subversion commands:

• svn commit - commit changes to the repository
• svn revert - revert local changes you have made
• svn add - add a new file to the repository
• svn delete - delete a file from the repository
• svn update - get changes from the repository

When working in teams, please do not email or scp files back and forth - use the repository instead! (That's the whole point.) Subversion keeps a full record of every version of every file (even deleted files), so keep in mind that you can always undo changes made accidentally.

Whether you are working alone or in a team, please remember to svn commit your changes regularly. If nothing else, subversion will at least serve as a file backup in case your hard drive fails. Also, don't forget to svn add any new files you introduce (and remember to update CMakeLists.txt too)

Submitting your code. Included in the code distribution is a shell script, submit.sh, that you will use to submit your code. The basic usage is "submit.sh n" where n is the number of the project component you are submitting. It must be run from the directory containing your code. After submitting your code, the script will also run a test build in a temporary directory to make sure that it compiles from the command line. Type "submit.sh -?" to see further information:

    [green@pc12 ~] ./submit.sh -?
    Usage: submit.sh <hw#>
    where <hw#> is a number in the range [1,5]

    Submits your project component by tagging the current version of
    your subversion repository as the submitted version.  It may be
    executed multiple times for the same <hw#>.  The most
    recently submitted version is the one that will be used for
    grading (and its timestamp will determine any late penalties).
    This script must be run from your subversion DavisDB directory.

    After submitting, the script will also run a test build of your
    project, by checking out the submitted version into a temporary
    directory and executing "cmake ." then "make".

    Note that since the submission procedure uses svn, if you are
    running submit from outside CSIF, it may prompt several times for
    your CSIF password.

    DavisDB%

• The due dates for all project parts are on Sundays at 11:59pm. For project parts submitted after 11:59pm on the due date, there is a 5% penalty applied to that project part for each hour late. Thus, a project part submitted 20 hours late will receive no credit. This policy holds also for the project extension proposal.

• Since emergencies do occasionally arise, each team is allocated 5 free late days for the entire course with no penalty. Once the 5 free late days have been used, penalties begin to be applied as described above.

Acknowledgments