inserting a row of data in a huge table.....

[alanchinese]

i have a talbe that contains 60M to 100M rows of data
column name.....data type.....length
...key..................char...........42
...desc................varchar......200
...ref...................int...............4

all three columns are indexed and key is the primary key.

i am doing the following operation:
if (!DataIsInTable(key))
InsertIntoTable(key);

its fairly slow when the number of rows are 100M.
i wonder if there are some advanced method to improve its performance?
what happen if i dont check if the data is in the table? would the failing process faster than selecting?
thankx....

 

[kejpa]

What database?
If youre using mySQL theres a REPLACE command that inserts if not present and updates (actually deletes and updates) if present.

Generally I sould say trying to insert and failing is faster than checking and then inserting. If you insert and the key is present then the insert fails, no harm done.

HTH
/Kejpa