Enterprise Application Integration (EAI) products provide
a strong platform for messaging and workflow management. However,
these products do not claim to specialize in either metadata management
– managing the underlying data resources and message formats that
participate in the EAI process – or data transformation – understanding
data and deriving accurate data transformation scripts for execution
within EAI environments.
As EAI projects scale, three problems arise if metadata
management and data mapping are not addressed:
-
Productivity: The ability to maintain point-to-point
data transformations becomes an enormous IT burden.
-
Agility: It becomes prohibitive to make
change in underlying data sources, message formats and business
rules. Critical business logic and metadata is trapped in
proprietary EAI products and versions.
-
Quality: As data is integrated without metadata
management, inconsistencies and ambiguities accumulate, resulting
in reduced quality of business information.
Metadata Management for EAI
A successful EAI deployment relies on three things: (1) a good
understanding of underlying data sources and application APIs,
(2) an awareness of any change, and (3) comprehensive management
of message formats. Inter-enterprise integration also requires
tracking partners’ data formats and industry-standard message
schemas. Proper metadata management is key to efficiently using
a complex EAI deployment. Metadata is best managed outside the
EAI product in a specialized metadata environment, as these are
more open, thereby reducing dependence on specific EAI vendors
and versions. This also allows the same metadata to be reused
by other users of the data resources such as application developers
and ETL/business intelligence infrastructure.
Data Transformation for EAI
EAI products allow messages to be physically transmitted through
a hub. But the different data sources and message formats are
often mapped point-to-point. Even when canonical message formats
are used, there are usually many different standardized messages
- with their versions changing over time - resulting in a spaghetti
of mappings that is difficult to maintain. SIM
allows every data source and message format version to be mapped
just once to the Information Model. All the end-to-end transformations
are then generated automatically. When a data source or message
format is modified, only one mapping requires updating and all
transformations are regenerated automatically.
Integrating Information Management & EAI Environments
The SIM environment easily and efficiently supports EAI run-time
environments with:
-
Transformation Code: May be based on a standard
such as XSLT, using code which is integrated to the EAI API,
typically using Java Beans, or in an EAI vendor-specific transformation
scripting.
-
Push/Pull: The SIM environment may push
translation scripts into the EAI repository or the EAI repository
may be configured using its API to request translation scripts
from the SIM environment, e.g. using a web service.
-
Design-Time/Run-Time: In high-volume systems,
it is not recommended that transformation scripts be generated
for each transaction. Instead, for each process a stable transformation
script may be generated and updated only when one of the data
sources, message formats or business rules changes.
The optimal choice will depend on both transaction
volume and the architecture and APIs of the specific EAI system,
such as IBM WebSphere MQ, Tibco, BEA, WebMethods, SeeBeyond, MS
BizTalk, Vitria, or Mercator.
The Benefits
Enterprises are turning to SIM as a major component of their integration
(EAI) infrastructure, recognizing that only the combination of
an industrial-strength metadata management system with next-generation
semantic based data integration capabilities will enable them
to:
-
Increase engineer productivity and code accuracy
by automatically generating translation scripts.
Reduce deployment time of integration projects.
-
Ensure consistency and maintainability by generating
canonical data formats from a single Information Model.
Accelerate adoption and tracking versions of industry standard
message formats (e.g. HL7, XBRL, ISO 15022, ACORD, and RosettaNet).
-
Replace the overhead and rigidity of manual point-to-point
mappings with the paradigm of mapping once.
-
Store business logic and rules in a central and
open way, thereby increasing business agility and reducing
dependence on proprietary EAI vendors and products.
-
Leverage a mature data management environment to
enforce policy, promote reuse, rationalize data assets and
ensure consistency as the project scales.
To request a Unicorn System
Solutions Sheet entitled, "SIM for Integration & EAI projects,
visit our
Literature page.