Performance: View vs. Subselect

Question

A little background:

I have a Log-Table with much information in it (around 10.000 entrys per week of using my webaaplication). I have a position-table, which is my main table (so the user works with positions in my application and can manipulate them etc.)
Now I want a LastStateChangedDateTime, so the DateTime of a event which I have in Log.

Now I can do this per 2 ways:

1) Per view.

I build a view in which I have the simple fields PositionID and LastStateChangedDateTime:

Select PositionID, Max(DateTime) as LastStateChangedTime from Position 
join Log on CAST(Position.PositionID as NVARCHAR) = Log.Message
where Event = 'PosStateChanged' 
group by PositionID 

And can connect the view in my which select:

Select bla, MyView.DateTime 
from Positions [Much more joins here] 
     inner join MyView 
     on Positions.PositionID = MyView.PositionID

Or

2) Per Subselect, which will be like:

Select bla, LastChangedDateTime 
from Positions [Much more joins here] 
    inner join (Select PositionID, Max(DateTime) as LastStateChangedTime 
                from Position 
                join Log on CAST(Position.PositionID as NVARCHAR) = Log.Message
                where Event = 'PosStateChanged' AND PositionID = Positions.PositionID
                group by PositionID) etc.etc.

So, simple question: What of both ways should I go and why? What is faster and why?

Answer 1

The statements are equal. A view is just the definition of a query, a placeholder so to say. When you use it in another Statement, the view's name gets replaced with the actual statement. So it is about readability and convenience and not about speed.

Some dbms offer special views, however, that store the actual query result. Oracle calls these materialized views. The idea is that if tables' data is rather constant for a long time, then why use the same complex query again and again. But then one must think of when to update the view.

But as said, normal views are simply names for pre-written SQL.