For common data sources check out this link which will give you some use cases and confirmation of support.
QlikView Data Sources | Qlik
With regards to Composite, the last time i worked with Composite Information Server I was using Cognos and at the time Composite published an ODBC driver that other tools can leverage to query the content in CIS. If still available ODBC is likely the easiest method to connect from QlikView since Qlik supports ODBC out of box.
In the link above you will see support for hadoop (check cloudera, hortonworks etc...) Oracle is also common , typically ODBC/OLDEB data soruces but there are many flavors of Oracle. Note that Qlik released a hyperion connector (also oracle) recently.
Qlik does full ETL to combine , ingest and integrate multiple sources of data into unified data models that provide excellent performance for analysis and data discovery. This is far more powerful and flexible than 'data blending' which typically refers to some mix of query federation and light etl. There is absolutely no limitation on the # of data sources , different types that you can model together in QLIK. The users queries are taken off the source data sources relieving them of what would be a performance nightmare . Most data models in QLIK run around 1-50 million records (ballpark). Outliers will be over 100 million records in which case you'll want a server to host the data model centrally to maintain speed for the users.
