|
|
Introduction
The Meta Integration® Works (MIW)
and its Repository (MIR) provide MITI's
data movement solutions. This MIW/MIR toolset is a development environment
generating data movement software components, also known as data bridges (executable code
that moves data from a source to a destination database/application) solving
three different categories of business problems:
- Enterprise Application Integration (EAI)
in which case both source and target applications are current systems that
usually complement each other and require periodic data updates. The data
movement can be enacted in real time (triggers), overnight (batch) updates,
replication, etc.
- Legacy Data Migration
in which case the source database is a legacy system that will be shutdown
after the data movement. This situation happens when migrating to a new
database delivered with a new application, or as part of the database
re-engineering of the same application
- Data Warehousing in
which case the source databases are under corporate business applications
and the target database is in fact your corporate data warehouse.
In the past, Data Movement has been solved by developing
specific software from scratch, in other words by programming data bridges
"by hand". This solution is obviously time consuming and not cost
effective. In addition, such data bridges are expensive to maintain with the
evolution (version upgrades) of both the source and target databases. However,
most data/application integration problems are still solved this way by most
consultants today.
Software Development Environments, specialized in the
development of Data Movement utilities, are now Commercially available
Off-The-Shelf (COTS). These products usually do an excellent job, however they
are primarily focused in solving specific data movements:
- Data Movements for data warehousing, or legacy data
migration only: very few of them are targeting the full problem of
data/application integration. One of the main reasons is the complexity of
writing to the Business Application Programming Interfaces (BAPI), and the
lack of standards in these interfaces. Note that data warehousing products
read from business applications, and write only in their own databases (data
warehouse).
- Data movements for the first generation of business
application (file based on mainframes) or second generation (relational
databases on client/server architectures): very few of them are targeting
the new generation of business applications based on 3-tier architectures.
Note that 3-tier architectures are now adopted by most business applications
in any vertical market (e.g. PeopleSoft, SAP, Baan, in the ERP market).
There are multiple benefits to 3-tier architectures, including portability
to multiple databases, the application servers can check integrity (business
rules) of data before writing into the database servers, etc.
MITI provides a solution that complements these COTS
products, by focussing on business data/application integration with support for
the new generation of 3-tier architectures. For example, the MIW's Model
Mapper allows the user to map:
- the Data Model of a RDBMS for a client/server based
business application, and
- the Object Model of the Business Application
Programming Interface (BAPI) for a 3-tier architecture based business
application. Note that the BAPI is not necessary in an Object Oriented
programming language like C++ or Java, in fact most of them are in C for
portability and integration with MS Visual Basic reasons.
Then, the MIW's Data Bridge Builder generates C++ code
which reads in the database (via ODBC) from the source business
application, and writes (via the BAPI) to the application server of the target
business application.
MITI also provides a solution exchanging data between a database and XML.
For example, the MIW's Model Mapper allows the user to map:
- the Schema/DTD of an XML file, and
- the Data Model of a RDBMS for a client/server based
business application.
Then, the MIW's Data Bridge Builder generates C++ code
which reads in the XML file (via an XML Parser) and writes into
the destination database (via ODBC).
Most data movement solutions offered by Data Warehousing (DW)
or Extract, Transform & Loading (ETL) products
make use of scripting languages often derived from the Basic language.
This technology presents several drawbacks including:
- Such scripting languages are derived from Basic, but are still different from Visual Basic.
Therefore, there is a learning curve.
- Data movement applications built with interpreted scripting languages (e.g. Basic)
are much slower than the ones built with compiled native languages like C/C++.
- Data movement applications built with scripting languages (e.g. Basic)
often rely on so called data movement server (interpreting the data transformations).
Therefore, it requires more system administration.
- These data movement servers and/or libraries often involve costly runtime fees.
One of the key feature that differentiates Meta Integration's data movement solutions
from its competitors is based on the generation of real native C++ code for Windows9x/NT
or Unix (Solaris, etc.). The data movement C++ code you generate is not only faster,
it is also truly yours with no runtime fee! Furthermore, you can manage and integrate
the generated code with your products and corporate infrastructure the way you want.
The data movement solutions produced by Meta Integration® Works (MIW)
can be exploited by several different types of users for different purposes.
- System Administrators, Data Integrators, or Data Warehouse Managers
can use the produced data movement solutions for real-time or batch updates
in multiple business contexts such as Legacy Data Migration (LDM),
Data Warehousing (DW), or Enterprise Application Integration (EAI).
In each case, the produced data movement solutions can be automatically
launched by your operating system scripts (Unix shell, Windows DOS Batch files),
or whatever corporate process enactment engine you might have. MIW generates
the data bridges, but does not provide you with a process enactment engine
for running the right data bridge, at the right time, and between
the right data stores or business applications. Such activities
are better left to system administrators by means of operating system features,
or by use of specific software products for all your existing corporate business process modeling
and enactment needs.
- Programmers can use MIW to produce "Data Movement Software Components"
which are completely integrated with your software applications in order to implement
the real time updates or import/export functionalities. In such case,
the produced data movement C++ code can be independently compiled and bundled
as extra dynamic libraries within your product delivery. In fact,
the data movement C++ code can also be integrated within your existing
version & configuration management solutions and go through your regular QA processes.
- End Users can also directly exploit the produced data movement solutions.
MIW includes a Visual Data Bridge tool (for Windows 9x/NT) customizable
with your company/product logo, and ready to package & deliver
to your end users or customers for their legacy data migration
or application integration problems.
Not only MITI's data movement solution integrates business applications based on
modern 2-tier and 3-tier architectures but MITI's Development Environment has itself
been implemented as a modern 3-tier architecture with:
- A DATABASE SERVER among the
most popular (ODBC enabled) RDBMS running on popular platforms in order to
adjust to IT departments existing technologies:
- MIR Personal Edition on Microsoft Windows 9X/Me/NT/2000/XP with Access97/2000
- MIR Enterprise Edition on Microsoft NT/2000/XP with SQL Server 6.5 to 2000 or on Sun Solaris 2.5 with Oracle 7 or 8.
- The APPLICATION SERVER is known as the
Meta Integration® Repository (MIR). MIR is an extensible application server
with both C++ & Java
Meta Integration® Software Development Kits (MIRSDK).
- The CLIENT is known as
Meta Integration® Works (MIW). MIW is a Web enabled client entirely written in Java
(100% pure Java)
The MIW toolset includes a model manager, model browser,
model converter, model comparator, model integrator, model mapper, data query, and a data
bridge builder. The model converter functionality of MIW is also available as a
Windows 9X/Me/NT/2000/XP standalone utility called
Meta Integration® Model Bridge (MIMB).
The MIR fully implements and integrates design methodologies
and metamodels from multiple standards. IDEF1X is both the military and industry
standard for data modeling (e.g. design of RDBMS). IDEF0 is also both the
military and industry standard for (high level) activity modeling required in
Business Process Re-engineering (BPR) and Improvement (BPI). IDEF3 (Process
Modeling) and IDEF4 (Object Modeling) are under development, these methods are
mission critical for global modeling of business processes.
The Unified Modeling Language (UML) is becoming the standard for object modeling,
now renforced by the arrival of the XML Metadata Interchange (XMI) standard
also managed by the Object Managemnt Group (OMG). Older standardization groups like
the CASE Data Interchange Format (CDIF) and the Metadata Coalition
are also taken into account.
This tool supplies several functionalities that allow you to
manage your metadata:
Once the metadata has been placed in the system you may then compare
two models. After having compared two models you may then decide to either
generate mapping information or to
integrate the two models.
The figure below describes the essential processes and flows of information between
the different tools of Meta Integration® Works and Repository.
Once the datastore models have been imported you may then specify how you
are going to migrate data between the source and destination datastores. The first step
before starting the migration should be to compare the models to see
how they are similar. The result of this comparison can then be used
as a basis for the data migration.
|
