and Performance Analysis Laureando Valerio Barbagallo Relatore Paolo Merialdo Correlatore Michele Aiello Context A database is a structured collection of data A database management system ID: 795239
Download The PPT/PDF document "In-Memory Database Competitive Landscape" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
In-Memory Database
Competitive LandscapeandPerformance Analysis
Laureando:
Valerio Barbagallo
Relatore:Paolo Merialdo
Correlatore:
Michele Aiello
Slide2Context
A
database
is a structured collection of data
A
database management system
(
DBMS
) is a computer software that manages databases
An
in-memory database (IMDB) is a database management system that primarily relies on main memory
DBMS
Database
Slide3IMDB: Application Scenarios
Real time applications with no durability needs (e.g.: router)
Real time applications with durability needs which require high throughput and low latencyTraditional applications during the development and testing phase
Slide4Objectives
IMDBs landscape and in-depth analysisMethodology for evaluating the database’s performance
Design and development of a benchmarkTesting and results’ analysis of different IMDBs
Slide5Summary
IMDBs vs. Traditional DBMSsDatabases’ Performance Analysis and Benchmark
Real Time Prepaid System Load Test CaseResults’ AnalysisConclusion
Slide6Comparison Against Traditional DBMS
Caching
Data-transfer overheadTransaction processing
Application
File System
Database Runtime
Database Cache
File System Cache
Hard Disk
Slide7Adding Durability
On-line backupHigh availability implementation
Non Volatile RAMTransaction logging
Performance
Durability
Slide8Advantages & Disadvantages
Advantages:
LightweightRobustnessHigh performance
Disadvantages:
Huge amount of RAM
Durability
Slow startup
Slow snapshot
Slide9Databases’ Performance Analysis
Databases are only slower or faster given a specific set of criteria in a given benchmark
Execution platformApplication scenarioImplementationTest Suite:Base test case
Load test case
Slide10Benchmark Requirements
A benchmark is a point of reference used to measure the performance
Requirements:PortableFlexibleDetailed report for each testVisual report
For both relational and object databaseEasy to use
Slide11Functional View
bench-properties.xml
benchmark.pdf
databases
tests
graphs
Slide12Real Time
Prepaid System Load Test Case
Real Time Prepaid System
Balance
check
Service authorization and management
Accounts management
10
tps
10 tps
2000
tps
Domain object
Initialization number
Account
4
millions
MSISDN
8
millions
Session
100
thousands
Slide13Throughput Of Account Management Task
PICO4
Prevayler
H2
HSQLDB
Db4o
Slide14Throughput Of Service Management Task
PICO4
Prevayler
H2
HSQLDB
Db4o
Slide15Memory Usage
PICO4
Prevayler
H2
HSQLDB
Db4o
Slide16CPU Usage
PICO4
Prevayler
H2
HSQLDB
Db4o
Slide17Conclusion
Contributions:IMDB’s in-depth analysis
Performance analysis problemDatabase benchmark framework with powerful graphsResults’ analysisFuture Development:
New tests and databasesGraphical user interfaceMapping technology layer
Slide18Thanks
for your attention
Laureando:
Valerio Barbagallo
Relatore:Paolo Merialdo
Correlatore:
Michele Aiello
Slide19Elements’ Description
The benchmark runs several Test
on different DatabaseEach Test is composed of (concurrent) TaskBoth Task and Test use many
MonitorEach Test can use many Reporter
Slide20File Size
PICO4
Prevayler
H2
HSQLDB
Db4o
Slide21Real Time Prepaid System Configuration
Slide22Concurrent View
Slide23Test Suite
Base test case
Load test case: Real Time Prepaid System
Balance
check
Service authorization and management
Accounts management
different
kind
of
objects
single task
fixed
number
of
transactions