Data Management for Embedded Systems: A Cellbased Approach - PowerPoint PPT Presentation

Loading...

PPT – Data Management for Embedded Systems: A Cellbased Approach PowerPoint presentation | free to download - id: 138e9a-YjVkZ



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Data Management for Embedded Systems: A Cellbased Approach

Description:

Data Management for Embedded Systems: A Cell-based Approach. Syed Saif ur Rahman, Marko Rosenm ller, Norbert ... Check Berkeley DB, COMET, and - Cellular DBMS ... – PowerPoint PPT presentation

Number of Views:45
Avg rating:3.0/5.0
Slides: 31
Provided by: asra2
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Data Management for Embedded Systems: A Cellbased Approach


1
Data Management for Embedded Systems A
Cell-based Approach
  • Syed Saif ur Rahman, Marko Rosenmüller, Norbert
    Siegmund,
  • Sagar Sunkle, Gunter Saake
  • University of Magdeburg, Germany.
  • Sven Apel
  • University of Passau, Germany.

2
Agenda
  • Motivation
  • Vision and Goals
  • Background
  • Cellular DBMS
  • Cell
  • Types of Cells
  • Summary
  • Future Work

3
Motivation
  • Existing DBMS
  • Complex
  • Too many functionalities
  • Monolithic architecture
  • Tight coupling
  • Many unused functionalities
  • Generalized solutions
  • One Size Fits All

4
Motivation
  • Existing DBMS
  • Performance less predictable
  • The consistency of performance with the increase
    of functionality and the data growth is not
    certain
  • Developed for
  • Legacy Hardware and Applications
  • Not suitable for data-centric embedded systems

5
Motivation
6
Motivation
  • Embedded Systems
  • Resource scarcity
  • Hardware heterogeneity
  • Software heterogeneity
  • Operating System
  • Programming Languages
  • Assembly/C/C/Java
  • Hot debate
  • Which one is the best?

7
Motivation
  • Embedded Systems
  • Programming Languages
  • Reduced Features
  • Dynamic memory allocation?
  • Multi-threading?
  • Not all solutions feasible
  • Virtual functions?
  • Console I/O
  • Pseudo-standards
  • Static memory allocation
  • Function inlinings

8
Motivation
  • Embedded Systems
  • Sensor Networks
  • In-Network Query Processors
  • Cougar
  • Tiny DB
  • Target only Berkeley Mote
  • Materialization points
  • Data storage on node
  • Aggregator nodes
  • Temporary storage during aggregation

9
Vision and Goals
  • Simple DBMS
  • Diversified architecture
  • Components
  • Simple
  • Small (Limited optimal functionality)
  • Reusable
  • Simple interfaces and interactions
  • Customizable Data Management
  • Hardware
  • Application
  • Data
  • User, …

10
Vision and Goals
  • Consistent performance
  • Highly predictable behavior with functionality
    and data growth
  • Efficient resource consumption
  • Reduced cost

11
Background
  • Embedded Database
  • Small footprint
  • Small set of tasks
  • Little maintenance
  • Multiple-platforms support
  • API based access
  • Check Berkeley DB, COMET, and -gt

12
Background
  • FAME-DBMS (http//fame-dbms.org/)
  • A customizable DBMS for embedded system
  • Software Product Line (SPL)
  • Approach to produce a family of related programs
    for a domain
  • High degree of customizability
  • Suitable for embedded systems (Leich et al.)
  • Feature-oriented Programming (FOP)
  • Paradigm for developing software product lines
  • Programs synthesis by composing features

13
Background
14
Cellular DBMS
  • A Cellular DBMS is composed of multiple cells.
  • Each cell is an atomic and autonomic instance of
    a customizable embedded database.
  • Could be a different SPL Variant
  • DBMS behavior
  • Collections of behavior of all cells

15
DBMS Cell
  • Cell
  • RISC-style architecture
  • Simple
  • Limited (optimal) functionality
  • Handle key-value pair
  • With data growth
  • Induce more cells
  • Virtually like binary fission

16
Types of cells
17
Composite Cell
  • Composed of multiple similar or different cells
  • From different we mean a variant of SPL
  • Usage
  • Table structure
  • Large data handling

18
High-Level Composite Cell
  • Composite cell composed of multiple composite
    cells in hierarchical order
  • For large data management

19
Hybrid Cell
  • Vertical
  • Different cell composition at different levels of
    high-level composite cells

20
Hybrid Cell
  • Horizontal
  • Different type of cell for different data ranges
    in composite cell

21
Evolution
  • Run-time transformation of cell
  • Constructive transformation
  • Previous form becomes an atomic integral unit of
    new form

Evolve
Evolve
22
Autonomy
  • DBMS Capabilities
  • Monitor
  • Diagnose
  • Tune
  • For consistent performance

23
Autonomy
  • Large autonomous DBMS
  • Using multiple embedded database
  • Customizable
  • Atomic
  • Autonomous

24
Implementation
25
Implementation
26
Implementation
27
Summary
  • Proposed a Cellular DBMS architecture
  • DBMS composed of multiple cells
  • Each cell
  • Atomic
  • Autonomic
  • Embedded Database
  • Variant of an embedded database SPL
  • Customized
  • Embedded device
  • Type of data

28
Summary
  • Cell-based approach ensures
  • Simplified architecture
  • Reduced complexity
  • More predictable behavior
  • Efficient resource utilization
  • Improved performance

29
Future Work
  • Techniques to use differently composed cells
    while minimizing code replication
  • Current implementation of cell evolution is
    explicitly programmed
  • Implicit learning
  • Self- capabilities
  • Implementing hybrid cell, cell mobility, resource
    balancing, and self- capabilities

30
Questions
  • Cellular DBMS
  • Database Research Group
  • University of Magdeburg
About PowerShow.com