Using grep or perl -ne 'print if /<expr>/' is linear search, which is an operation that will scale with the size of the file. If the file is double the size, you can expect it to take twice as long time.
If you use a different storage format, such as dbm or the like, you can improve this performance significantly, but it has nothing with the way the file is being read - its all about the format of the file.
If you still want to keep the flat file you have described, you can build an index in a separate and use that for look up. One tool that will allow you to index the data is Berkeley DB for which there is the DB_File module.
| 