firoz -- "fastest" isn't always best. The only type of index limited to one per table is a clustered index. The table's data is physically stored in order of the clustered index. Now image this for example: You have an "Invoice" table with "CreateDate" column and have a clustered index against that column in descending order. Now your searches based on the CreateDate column will be fast, the inserts will be slow. When you insert a new record it will have a newer createdate and have to physically shove all of the records down a notch to write at the top portion of the file descriptor, giving you very slow inserts. Everything in computing is a cost-benefit tradeoff, indexes are no exception.
Why don't you explain what you're trying to accomplish or learn?