Assume you have a system you need to connect to but there is no ODBC driver for it, nothing. Or a WebService that has special requirement. In that case, the Adapter SDK is the way to go.
There are many kinds of applications that not necessarily look like a database system and thus do not provide an ODBC driver. Examples can be all kinds of EAI tools like IBM MQ, Tibco, JMS, or different internet applications. For the most common ones we have written Adapters ourselves already using the Adapter SDK.
You can even use Adapters do implement custom functions, but two things you cannot:
Write Transforms - an Adapter is row based whereas a transform requires data to be loaded first, then data gets transformed and finally sent to the output. But one row can contain a nested schema with many child rows, that might help often.
Access DataStores - An Adapter is a datastore itself, it cannot read/write another datastore's table.
And another requirement is key, you have to be able to contact your source system via Java, as the Adapter SDK itself is written in Java.
As said, the Adapter itself is like a DataStore, therefore when writing an Adapter some capabilities have to be implemented like the structure of the objects can be imported - DI requires tables, columns, schemes to work - and the Adapter operations like read from a table or similar have to be implemented.
- Integration of Adapters into DS
- Types of Adapters
- A simple Table Read Adapter
- Example II - the eMail Adapter
- Installing the Adapter