Do you have multiple partitions or just one? If you have multiple partitions then you should use Query Binding and each partition should have a WHERE clause that specifies its data. Multiple partitions reading the same data can lead to locking problems.

If you have only a single partition then let me know.

The data source shouldn't need anything other than read permissions (unless you have ROLAP or writeback). So, given the data source in the SSAS database, how is it logging in to SQL Server? Is it using a service account? If so, that probably has read/write. If it's a specific userid/password, make sure that userid only has read access to the data.

I'm still looking into why, even if it has read permissions, it woudl be placing blocking locks on the tables that make up the views.

Now, to get on a soapbox for a moment... why are you accessing views that hit the OLTP tables in the first place? It sounds like the dimensional model was done in views instead of creating physical tables and using an ETL process to copy the data. Is that correct? One reason for building a star/snowflake schema is to avoid loading down your OLTP with the heavy queries involved in processing.