You should consider using Hadoop if:
You are dealing with large amounts of data per day (terabytes and up).
Other tools (relational databases, key-value stores) are not able to keep up with your storage or processing requirements.
You've already scaled up your server hardware to the point where buying the next bigger server is very expensive compared with buying a lot of cheap servers.
Other tools for horizontal scaling are too expensive, inconvenient, or don't address important aspects of your problem as well as Hadoop does.
You should consider using Hive if:
You don't want to write map-reduce jobs in Java.
Your data analysis needs map neatly to SQL-like queries on your source data.
You want to support exploratory queries performed by people who know more about data analysis than about writing map-reduce code.