I don't think that feature existed in 5.5.38:
"
----- 2012-08-07 5.6.6 Milestone 9 -- Functionality Added or Changed -- InnoDB -----
InnoDB now supports the DATA DIRECTORY='directory' clause of the CREATE TABLE (
http://dev.mysql.com/doc/refman/5.6/en/create-table.html ) statement, which allows you to create InnoDB file-per-table (
http://dev.mysql.com/doc/refman/5.6/en/glossary.html#glos_file_per_table ) tablespaces (.ibd files) in a location outside the MySQL data directory.
"
Check the 5.5 manual page to see if it explicitly mentions InnoDB. The DATA DIRECTORY feature was implemented eons ago for MyISAM, but only recently added to InnoDB and PARTITIONing.
> table across multiple disks to get the space and improve throughput
You are unlikely to get any benefit. Why? Because operations will tend to be serial -- moving from one disk to the next, getting little or no overlap.
You get much more bang for your buck by striping the disks. Production machines standardly use RAID-5 for both striping and fault tolerance.
By using striping, you have a better chance of making use of multiple heads, even within a given partition.
> char(30)
This is a bad choice unless the field is always 30 characters. Even so, if it is utf8 and not an Asian language, CHAR(30) is not as good as VARCHAR(30).
> seq text(1000) NOT NULL DEFAULT '',
100M rows with TEXT? Is that field repeated any? Would normalization help?
Or is the text usually 'long'? If so, consider compressing it before storing it. (And then store it in a BLOB.) English text generally compresses 3:1.
How fast are you inserting? You have 3 secondary keys; each needs to be updated eventually. This can be costly during the INSERTs.
How much RAM do you have? Setting innodb_buffer_pool to about 70% of that is the most important tuning to do.
You mention "throughput"; what are the common SELECTs?