Recently I noticed a lot of parallelized queries (create table as select...) which cause high library cache lock concurrancy.
In the following question Mihail stated out to beware of library cache locks in long running CTAS in the form
create table <new_table_name> parallel <partitioning_info> as select * from <old_table_name> where <filter>;
Faster way to load huge data warehouse table
So why is this? The ratio of hard parsing is very low. Is this a problem because all the sessions try to look up the execution plan in the library cache? I thought by soft parsing there is only a pin on the library cache object?