HiveQL's similarity to SQL is both a benefit and a hazard.
On the one hand, HiveQL supports a lot of familiar operations: SELECT, GROUP BY, LIMIT, UNION, some support for subqueries.
On the other hand...or plethora of hands...
Hive can replace the contents of a table wholesale with INSERT OVERWRITE TABLE, but cannot selectively UPDATE, INSERT, or DELETE its contents.
Hive is best understood as a layer over an HDFS data store and a way to query its contents, not as random write access to that storage mechanism.
Hive also doesn't have transactions. Again, it's not a random write access data storage layer.
Only one table or view is supported in a FROM clause.
Subqueries are only allowed in a WHERE clause. Correlated subqueries are right out.
Hive does not support materialized views.
Not all SQL-92 data types and functions are supported.
There's no expectation of sub-second latency in Hive queries. Some queries could take minutes or hours. Almost none will be shorter than several seconds.
There are more.
If you keep these distinctions in mind, you'll be fine. Otherwise, you may spend some time being confused.