Ei dian otsikkoa

1
Database comparison model
Jari Porrasmaa University of Kuopio, computing
centre HIS research and development unit
2
Topics for today

• Conversion model - From FileMan to objects
• main principles
• standards and technologies used
• Other parts of the comparison model
• a list of comparison criteria
• modifications (add/delete/change) to this list

3
Very short background

• M(umps) is a programming language that offers
  automatic persistence for data structures
• FileMan (FM) is a DBMS based on Mumps
• originates from USA (VA,DOD), but is also used in
  Finland, Germany, African countries,...
• FixIT is a RAD development toolkit that enables
  building of GUIs for FileMan databases
• Windows apps ( delphi FixIT)
• Java apps (client applets or server beans, Web
  FixIT)
• HTML apps (front ends, e-FixIT)
• Please refer to other presentations and project
  web pages for more background material

4
Introduction, what and why?

• What? Comparison of different database types
  using one example case (the demo application,
  Labworks). Also database features as is with the
  demo app.
• Why? To research utilisation of component based
  software engineering paradigm in the healthcare
  domain. Also to provide a strategy for
  introducing new technology into existing software
  assets in case
• current technology (FM) dies and needs to be
  replaced
• new technology gives enough additional value
• It is NOT a beauty contest! No one clear winner
  in the end. The objective is to bring out the
  strengths and weaknesses of each solution in
  context of different starting points in client
  organisations (hospitals and companies in the
  healthcare software business).

5
Contents of the database comparison

• How can the Labworks demo app be implemented with
  DBMS (and related) technology family X. Work done
  by vendors (them) and HIS RD (us) together.
• Each vendor presents their own solution,
  specification of what is actually implemented
  together
• Conversion of the legacy database into each
  target database (mainly on paper, but maybe
  prototypes?)
• Evaluation of other database features (the stuff
  generally associated with DBMSs)
• Comparison can be thought as a model for other
  comparisons (hospitals and companies can attach
  other technologies and products to it)

6
Databases to be compared

• Representative of each DBSM genre
• Cache Objects
• roots in M language, has OO extensions
• Oracle
• based on the relational model, has OO extension
• Jasmine ODB
• pure object model representative
• FileMan
• viability of legacy technology, starting point
  for other implementations

7
Conversion model

• Its goal is to represent FileMan datamodel, data
  and functionality as an object oriented model
• Also to find a standard object model(s) which can
  be used in the conversion process

8
Conversion model in more detail

• Mapping of FM data model into OO data model
• FileMan stuctures
• FileMan datatypes (fields)
• automatic and manual conversion (refactoring the
  generated model, or mapping FM to a non-generated
  model)
• Conversion of FM data into objects
• Implementation of FM functionality in other DBMS
  technologies
• Selection of the object model used in the
  conversion
• Possibly specification of gateways and wrappers
  to integrate different technologies

9
High level view of conversion and comparison
source and targets
Jasmine (ODQL)
FileMan data definitions and data
Cache (CDL)
Oracle PL/SQL?
10
Help from the ODMG standard?
Jasmine ODQL
ODGM standard
FileMan data definitions and data
Cache CDL
Oracle PL/SQL
11
Help from SQL1999 standard?
Jasmine ODQL
FileMan data definitions and data
Cache CDL
SQL1999 (SQL3)
Oracle PL/SQL
12
Metadata can be utilised more generally...
Jasmine ODQL
CWM
FileMan data definitions and data
Cache CDL tiedosto
SQL1999 (SQL3)
Oracle PL/SQL
XMI
FixIT (DoIT, etc.IT) toolkits interoperability
with other software development tools
etc. etc
XMI CWM explained soon...
Component integration
Modelling tools
13
The overall picture...

• Using the model for other purposes than the
  conversion
• Modelling, other developement tools, FixIT
  toolkit, etc.

14
OMG MOF, XMI and CWM explained

• OMG open consortium producing spefications/standar
  ds based on requests made by members
• known mainly from the following specs CORBA
  UML
• MOF - Meta Object Facility, unifying MOF
  compatible metamodels (MOF is a metametamodel)
• XMI - XML Metadata Interchange, Interchange of
  metamodels using XML
• CWM - Common (data)Warehouse Metamodel,
  application of XMI (and MOF and UML) for
  datawarehouse model and data interchange

15
XML Metadata Interchange (XMI)

• Generates an XML structure definition (DTD) from
  a MOF compliant metamodel (eg. MOF metamodel for
  UML generates the UML XMI DTD)
• DTD can be extended!
• Generates an XML document from metadata (metadata
  is an instance of metamodel, eg. UML model
  generates an XML document)
• Can be seen as a MOF interchange format or as an
  independent metadatastandard
• Meta layers are not absolute. XMI can be used for
  data interchange also.

16
Simplified XMI as a diagram
Metamodel layer
If one uses a UML model on the metamodel layer
XMI can then be used as a data interchange format!
model layer
17
Common Warehouse Metamodel (CWM)

• Defines the CWM MOF metamodel which is then used
  for generating the interchange format
• Generates from the MOF CWM metamodel the XMI CWM
  DTD
• Generates from a MOF CWM model an XMI CWM
  document
• Metamodels can also be interchanged as documents
• Defines APIs for handling datawarehouses and
  metadata
• Based on UML and XMI
• Could it be used in conversions by stretching the
  ELT scenario specified in CWM standard?

18
Usage scenarios for CWM

• Taken from the CWM specification
• ETL Extract, Transform and Load
• Data is extracted from some sources, then
  transformed into the desired format and finally
  loaded into some destination. Types of sources
  and destinations can vary. Transformations can be
  performed in different stages and several times.
• Could it be used for DB 2 DB conversions?
• Which vendors will support CWM?
• CWM specification name many other usages for CWM,
  but these are not of any interest here

19
SQL1999 (SQL3) and basic-SQL

• SQL-99 is a very broad standard which introduces
  many OO concepts into the SQL world
• Implementations now and in the future? (even
  partial implementations on the OO part of the
  standard could be very useful)
• Wide implementation of Basic-SQL is a strong
  plus for using SQL, also SQL standard family is
  the strongest one in the DBMS world
• Significance of other SQL standards like SQLJ?

20
Other standards?

• Are there other standards that weve overlooked
  and could be useful for us?
• In regard of the datamodel and data conversion
• and in the integration of development tools
  trough the use of metadata

21
Using standards pros and cons

• Are the tools and software supporting standards?
• Amount of work and new things to study
• Which actual products support different standards
  and to what extent? Future plans?
• Standards can help in achieving some degree of
  interoperability, integration, product
  independence etc.

22
FileMan structures as OO structures
23
FileMan field types and other features in the OO
model
This array doesnt have all FileMan features
(e.g. printing, user privileges). More complete
description in a separate document!
24
Other comparison criteria

• Support for object orientation
• Support for development tools
• Support for component architectures and standards
• General DBMS features
• Market situation of technology, costs, vendor
  situation on the market
• Each criteria will be assessed according to
  resources available and importance of criteria

25
Object orientation

• Software engineering is done in an object
  oriented way. The persistence layer should also
  support OO.
• DBMS implements concepts generally associated
  with OO
• DBMS offers bindings to OO languages
• DBMS has support for OO modelling and analysis
  software

26
Component architectures

• Comparison is architecture independent so
  databases should interoperate with several
  architectures
• CORBA
• Windows DNA (.NET)
• Java 2 Enterprise Edition
• Can DBMS provide persistence for components
  implemented in different architectures?
• Also a separate master thesis on different
  products implementing architectures mentioned
  above (Juha Rannanheimo)

27
Support for standards

• Standards on a more general level (not just the
  standards used in the project)
• Standards on database languages
• Database interface standards
• Standards on transaction control
• Metadata standards

28
Development tools

• For each tier of the target architecture
• presentation, user interfaces (although focus
  mainly on the layers below)
• enterprise, business logic
• persistence, DBMS
• Reporting tools
• How do different tools work with a particular DB?

29
General database features

• Hardware and operating system platforms
• User management (on DB and application level?)
• Authorisation
• Concurrency control
• Security
• Backups
• Administration tools
• ...

30
Performance and scalability

• No resources for doing benchmarks. Any applicable
  and independent benchmarks available?
• Experiences and observations from prototypes
• Performance conversion / production
• Scalability up/down, same application running in
  small and very large environments

31
Vendor and technology specifics,soft issues

• Vendor and technology situation in the current
  market / future?
• Independent evaluations?
• soft issues
• available human resources?
• learning curve of technology?

32
Your comments / ideas?

• Was there something missing?
• Something too much?
• What is essential?
• What do you as vendors want to tell to people
  reading the report?