Please enable JavaScript.
Coggle requires JavaScript to display documents.
Introduction (Myths about DBs (9) (Tables should have one or two indexes,…
Introduction
Factors of database performance
Hardware
Processor speed
Number of processors
Number of disk drives and I/O bandwidth
Size of main memory
Communication network
Type of architecture
Software: type of database technology used
Database tuning
Indexing
Maintained views
Join indexes
Data duplication
Improve database peformance
Apply skilled techniques known to work
Understand underlying architecture
Research and benchmarking
Tools to monitor and tune the performance of the system
Train the application, database and system personnel
Myths about DBs (9)
Logical design should always map exactly to physical design
Not true
Physical design mandated by performance: high speeds can be gained by appropriate physical design and indexing
Most of the processing cannot be performed in SQL
False
Inefficient to deal SQL as an I/O handler
ODBC: open database connectivity allows a standard software interface for accessing DBMS
Tables should have one or two indexes
False
Storage is not an issue
Computing is cheap
Should have as many indexes as needed to improve performance
Tables should not be too big
Not true with good indexing and good physical designs
All RDBs can now handle very large relations
Defaults are OK
Generally not true
Must select appropriate flafs (parameter setting) to improve performance
Never use negative clauses in SQL
False
Negative clauses performed with a positive clause; used as filters
Uncommitted read is a dirty word
Depends
Give getter performance in some application domains
There are no locking problems since there are no timeouts or deadlocks
False
Deadlocks can be disastrous, still need to monitor
Compression is always good
Depends on application
Generally true
Database Technologies
Simple file systems
Relational database systems
Object oriented database systems
Deductive database systems
Key-value pair based database systems
NoSQL
Basic Hardware
Storage Systems
RAID
Storage area networks
Formulas
Moore's law
Joy's law
Hit ratio
Disk access time
Transmit time
Database architectures
Centralised
Client server
Distributed
World Wide Web
Grid databases
P2P databases
Transaction Processing
Definition
ACID
Provided functionality or an application development
Features
Software
Processes
Threads
Message passing
Session-based
Datagrams
Scheduling
Goals
Maximise utilisation
Minimise response time
Conflicting goals
Scheduler should no push for very high utilisation