Best Practise for several index table

[ledy]

Hi,

The content, size and queries for table is known already. However, it's not clear yet, how to set it up best.
Even noSQL might be an option and maybe You are experienced with a similar project or have tips for a best practise scenario.

-table structure: identifier1 (sha1 hash), identifier2 (md5 hash), identifier3 (varchar(30)), 600 * tinyint, 8 * varchar(30), 1 * varchar(300)
-most popular read-query: SELECT * FROM this_table WHERE identifier1="..." OR identifier2="..." OR identifier3="...";
==> at least one of the identifier must fit. Only == matches, no "LIKE %"-queries and no other exhausting filters.
-number of rows: beetween 1`500`000`000 and 1`800`000`000.
-speed requirements for reads: as fast as possible, lots of queries per second need to be processed
-speed requirements for writes/updates: may be processed with little delay, no real time requirement

What can You recommend?

Thanks a lot!

[rmiao]

On which rdbms?

[ledy]

The server will only handle the task as described in the post before.
Even the (r)dbms is not chosen yet and tips are very welcome.

We could start out of the box with images incl. postresql, mysql/innodb, mysql/isam or with a bit more effort for setting up and modifying the API at the other end: mongoDB

What's fitting best for this case?

Thanks