question

I am considering using `hierarchyid` type as a part of clustered primary key. The key itself will be composed by the following types: - BIGINT - INT - HIERARCHYID in the given order. Generally (in 90% of the cases), the value of the `hierarchyid` column will be set to `/` - in other words there will be only one record for the given `BIGINT - INT ` value pair. I have read some similar articles/questions ([Clustered Index Design Guidelines][1], [SQL primary key: integer vs varchar ][2]), and generally the only thing I found is about the use of wide columns. I in the [official documentation][3] is said that the length of `hierarchyid` can be 5 bytes or more, but if you create a table you can see the `Length` is exactly 892 bytes. Is this mean that the size of the primary key is increased with `892` bytes always, or it is increased with the actual size of the column only (for example 5 bytes)? Also, if someone have experience with using the such column as primary key it will be grateful to share some experience with us? [1]: https://msdn.microsoft.com/en-us/library/ms190639.aspx [2]: http://stackoverflow.com/a/3162628/1080354 [3]: https://msdn.microsoft.com/en-us/library/bb677290.aspx