Database Tuning

Introduction
Description
Database ERD
Experiment Queries
Hardware Characteristics
Recommendations
References
Contributors
License

Introduction

In real world situations, creating database projects is more just making a scheme that fits the description. We are usually faced with a lot of questions like:

What DBMS should we use?
Is it better to make it relational or NOSQL?
How to get the best performance?
What Hardware spec. are recommended to use?

Description

In this Repository we experiment with the different concepts of database tuning and the different techniques used like:

Optimizing the schema.
Optimizing the memory requirements.
Optimizing the queries.
Testing the performace of MySQL vs MongoDB.

Database ERD

Experiment-Queries

Select ID, Name, and Gender of students who got an "A" in a course final exam. (Schema Optimization was applied)
Select all the research projects worked on by professors working in a certain department along with the professors’ names. (Cache was used)
Select ID, Name, Gender, and Salary of all professors with a salary less than 1200. (B-Tree Index was used)
Select ID, Name, Gender, and Salary of a certain professor using his name. (Hash Index was used)
Select ID, Name, Gender, and Salary of all professors with Certain Name whose Salary is greater than 6000. (Hash Index and B-Tree were used)
Select student ID, Name, and City of students with a specific name or who live in a specific city. (Query Optimization was used)

Experiment Results

MySQL Performance

MongoDB Performance

MySQL vs MongoDB Performance

Hardware Characteristics

OS: Windows 10 64-bits
CPU: Intel® Core™ i7-8750 CPU @2.20GHz (12 CPUs)
RAM: 16 GB
Hard disk: 1 TB HDD

Recommendations

Using indexing in MySQL improves the queries execution time significantly at the cost of additional memory and slower updates.
Taking your time to design the schema is a very important step in database design as we showed how schema optimization significantly improves the execution time of the queries involved in that optimization.
Using SQL is better than NOSQL in complex queries that join multiple tables together, while NOSQL is better in retrieving a huge number of records from the same collection so when designing the NOSQL schema we should consider Embedding rather than referencing whenever possible.
Using caching significantly decreases the execution time of frequently used selection queries. So we recommend using caching whenever possible but note that when data is too large you need a very large cache to have that significant speedup.