Generate Surrogate Keys: SQL 2012 Sequences vs Identity vs SSIS Generated

Hello,

Identity columns and SSIS generated row numbers (and GUIDs) were candidates for surrogate keys in dimension tables under SQL 2008.

SQL 2012 introduced a new feature titled Sequences, which are said to be faster than identity columns but the use of them may produce gaps in numbers produced. This is not an issue for surrogate keys.

Could someone point me to a comparison between these 3 options? When loading  dimension tables via SSIS, which of the above features is suitable? Is there a standard design pattern?

Thanks