DIALOG INTERFACE
FOR DYNAMIC DATA MODELS
Mgr. Vojtěch Přehnal
Presentation outline
• Introduction
• Field of research
• Motivation
• Objectives
• State of art
• Goals and gains
• Conclusions & Future work
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 2
Introduction
• New approach to information system development
• Aimed at:
• large-scale time-varying data models
• 3-tier client-server architecture
• relational DB
• Platform independent methodology
• various operating systems
• various application servers
• various database engines
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 3
Terminology
• Item: an ordered multi-set of related data values. The
rank of each data value is referred to as an ordinal
• Field: a multi-set of data values with the same ordinal and
some common attributes (type, size, nullability, identity,
…)
• Table: a multi-set of data values organized using a model
of items (rows, records) and fields (columns)
• Data model: a set of tables and their relations
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 4
Terminology
• Dynamic data model: data model with time-varying data
structure, fully editable at the run-time
• Dialog interface: communication protocol between client
and server
• Meta-data: supplementary information defining the
structure and the attributes of the raw data
• CRUD operations: standard database operations with
the items: Create (insert), Read (select), Update and
Delete.
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 5
Motivation
• Information system for ZONER software, a.s.
• 3 divisions (Internet Services, Software, Zoner Press)
• 15 branches world-wide
• over 120 empoyees
• over 350 tables with common CRUD operations
• only 8 tables with specialized logic (!)
• tens of millions of items
• over 40 GB of (textual) data
• average time to data-model adjustment:
only 14 hours ( ~ cca. 2 adjustments / day)
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 6
Objectives
• Enable editing data model at run-time
• Automate the most common practices
• Separate common and specialized logic
• Lower the costs of SW development
• Preserve performance, scalability and adaptability of
traditional approach
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 7
3-tier architecture
• means three-tier development
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 8
changechangechange
3-tier architecture
• How to automate changes in application and presentation
tier according to changes in DB?
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 9
changechangechange
State of art
• Compile-time development automatization
• automated source code generation:
• automated class model generation
• ORM tools (ADO.NET, Linq2Sql, Telerik OpenAccess, … )
• automated application logic generation
• Microsoft WCF RIA Services, Picasso, Habanero, …
• automated application and UI code generation
• Microsoft ASP.NET Dynamic Data, Microsoft WebMatrix
 auto-generated source code can be extented with specialized logic
arbitrarily
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 10
Automated source code generation
• Limitations:
 source code based on fixed data model  rebuild required for
every change in data-model
 common and specialized logic is mixed together  loss of
specialized business logic after re-generating source code
 performing uniform changes in common functionality for all the
tables requires rewriting source code for each individual table
 efficient only for systems with lower number of tables and with
majority of specialized functionality
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 11
Goals and gains
• Dialog interface1) for:
• interactive data-model exploration
• automated user interface generation
• data retrieval, validation and change submission (CRUD ops)
• automated data log creation
• automated SQL-injection protection
• centralized UAC (user access control) management
• data model and user interface localization
1) Dialog interface = communication protocol between client and server:
• data model of messages exchanged between client and server
• sequential model of these messages (their arrangement in time)
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 12
Conclusions
• New methodology for information system development
• Key point: retrieving data model at the runtime
• No brand-new revolutionary techniques
(instead, utilizing well-known techniques for new purposes)
• Main advantages:
 ability to edit data model on the fly
 no need of rebuilding or restarting the running application
 no need of rewriting source code of common operations for each
individual table
 suitable for the vast majority of data-driven apps
 easily extensible for individual user´s needs
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 13
Future work
• Study of various business scenarios
• Design of a meta-data model and a dialog interface to
pass the broadest possible scale of user requirements
• Simple application for demonstration purposes
• Presentation in science community
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 14
Discussion
• Comments and questions
Mgr. Vojtěch Přehnal Dialog interface for dynamic data models 15