CS 6244 - Database System Implementation Review

Introduction

In Spring 2025, I took CS 6422: Database System Implementation.
It was the first time the course was offered, so there were no reviews yet on OMSCentral or OMSCS Hub.

When I was planning my courses, CS 6400: Database Systems Concepts and Design didn’t seem like the right fit for me because it focuses on foundational topics I already know.
Database Systems Implementation, on the other hand, looked like a deeper, more technical course that filled a gap in the program.

Highlights

The course was centered around BuzzDB, an educational database project built by the instructor.
The highlight for me was the project on implementing a B+ Tree. I had never built one before, and I wanted to deepen my understanding of how it works, especially since B+ Trees are widely used in real-world database systems like PostgreSQL,and MySQL..

Other important topics included:

• Tree Structures: B+ Trees, Patricia Tries, Inverted Indexes, R-Trees, and Learned Indexes (a new concept for me).

• Compression Algorithms: Delta Encoding, Bit Packing, Huffman Coding, and Byte Dictionary Compression.

• Page Management: Understanding how databases handle pages, align page tables, and how settings in systems like PostgreSQL affect performance.

Programming assignments were posted well in advance, usually giving 3–4 weeks to complete each one, which made it easy to plan ahead.
Assignments were submitted via Gradescope with no hidden test cases, which I found to be a huge plus. You could start early, power through in a few days, and be confident you earned 100% long before the deadline.

Overall, it was a low-effort course that paired well with a harder class, while still offering practical insights into database internals.

There’s also a planned Part 2 course that will go deeper into advanced topics, but it might take a year or more before it becomes available.

Lectures

The lecture videos were clear, well-structured, and easy to follow.
Each module introduced a real-world database problem, then explained how BuzzDB approached it, often showing code snippets directly from the repository.

The material covered both fundamentals like storage management and more advanced topics like query execution and vectorized execution.

One small critique: I would have liked to see more examples from real-world systems like PostgreSQL, instead of focusing almost entirely on BuzzDB.

Topics covered:

• Storage Management

• Slotted Pages

• Buffer Management

• Multi-Threading

• Hash Tables

• Concurrency Control

• B+ Trees

• Other Tree Structures

• Query Execution

• Operator Frameworks

• Query Compilation

• Columnar Storage and Vectorized Execution

Projects

• Project 1: C++ Warm-up — read/write Excel files.

• Project 2: Build a basic Buffer Manager using a 2Q caching policy. Very easy.

• Project 3: Implement a B+ Tree. This was the most fun and most challenging — properly handling node splits can get tricky. Took around 10 hours.

• Project 4: Implement database Operators (SELECT, SORT, JOIN, UNION, etc.). Straightforward but a little tedious — around 20–30 minutes per operator, with ~12 operators in total.

It’s very achievable to get 100% on all projects, since the tests are provided upfront and you have 3–4 weeks to complete each assignment.

Conclusion

If you enjoy tougher OMSCS courses (like Distributed Computing, Advanced OS, or HPCA), this one might feel a bit too easy, especially if you're already comfortable with C++ and basic database concepts.

Still, I picked up a lot of practical knowledge, especially around internal database optimizations and newer indexing techniques.

For a first-time offering, the course was well-organized. There were a few minor issues, like deadlines being announced a little late, but nothing that made the course confusing or hard to manage.

Overall, it was a low-stress, practical, and informative course.
If you want a lighter semester while still learning useful things, this is a great one to pair with a heavier class.