index on url or hashing considering RAM

Question

I am working on a project which needs to add/update around 1 million urls daily. Some days are mostly updates and some days are mostly add and some days are mix.

So, on every query there is need to look up uniqueness of url in url table.

How look up for url can be made really fast because at the moment index is set at url column and it works good but in coming weeks RAM would not be enough if index are kept on same column and new records will be added in millions.

That's why I am looking for a solution so that when there will be 150+ million urls in total then its look up should be fast. I am thinking of creating indexing on md5 but then worries about collision chances. A friend tipped me to calculate crc32 hash also and concatenate with md5 to make collision possibility to zero and store it in binary(20) that way only 20 bytes would be taken as index instead of 255 currently varchar(255) set as url column data type.

Currently there are total around 50 million urls and with 8GB ram its working fine.

Yesterday, I asked a question url text compression (not shortening) and storing in mysql related to the same project.

[Edit] I have thought of another solution of putting crc32 hash only in decimal form to speed up look up. And at application level porting a check on how many records are returned. If more than 1 record is returned then exact url should also be matched. That way collision would also be avoided while keep low load on RAM and disk space by storing 4 bytes for each row instead of 20 bytes (md5+crc32). What you say?