SQL Server: is there a way to theoretically estimate retrieval time of a query

Question

i will be having a Database with more than 8 hundred billion records,

i am gonna be querying the DB via SQL Server ONLY to retrieve records i.e. SELECT queries only.

i am gonna be retrieving from 1 to 36,000,000 records, plus i will also think about caching (36,000,000 x 8 ) records.

what i was looking for is some means through which i could get an estimation of HOW MUCH TIME will it take for a simple SELECT query SQL Server (any version\edition) to get back 36,000,000(worst case scenario).

My table is simple with only 9 columns\attributes, all of them INT(integers).

I need to get an idea of this before hand before i could populate my DB & design my App while keeping in mind that whether i have to rely on SQL Server or my App has to rely on huge caching because slow response is NOT an option.

EXTRA NOTE

• Network Connection and bandwidth is not a consideration here.
• I cannot check elapsed time via programming, because DB is\can not populated yet.

---

E D I T

why the down vote

Answer 1

Populate your table with junk data and turn on statistics.

Answer 2

No, sorry, no easy silver bullet. You an deduct that from other queries if you have a sensible data reference set and can live with variations, but otherwise - no. People earn money tuning that ;)