Mirroring is relatively easy to setup and does not need any additional hardware to configure it. So mirroring is good to get started with when the application and database is small.

If you already have a clustering environment, you might want to to use it as your first choice. However, if you don't have the hardware resources readily available, you might want to decide between investing in buying the required hardware or go with mirroring.

Mirroring comes with some overhead. If you are using Mirroring with high safety mode, and if your database is highly transactional, you might get a performance penalty due to mirroring. So sometimes people prefer NOT to use mirroring when the database is big and highly transactional. Again, all these parameters are subjective to the specific environment and type of data being processed.

So to answer the question "when would i use mirroring instead of clustering", the immediate answer is "you will always use mirroring if you dont have clustering resources available". And a more generic answer is "you will start with mirroring when the database is relatively small and as it grows you might decide to go for clustering at a later stage"

Looking forward to hear more ideas from people around