in reply to MLDBM performance problem

Maybe you should consider switching to a relational database backend? MySQL, PostgreSQL, Oracle, IBM DB2, .. you have many to choose from.
--
print map{chr}unpack(q{A3}x24,q{074117115116032097110111116104101114032080101114108032104097099107101114})

Replies are listed 'Best First'.
Re^2: MLDBM performance problem
by henrywalker23 (Novice) on Jun 10, 2007 at 13:27 UTC
    Thanks for the suggestion, but all I need is a persistent hash where the values are array references. I'm worried that a database such as you suggest would involve a lot of overhead and be counter-productive as far as performance is concerned.
      I'm worried that a database such as you suggest would involve a lot of overhead and be counter-productive as far as performance is concerned.
      Perhaps you're right, but unless you Benchmark the two approaches, that's pure speculation. And, if you have issues with paging due to data not fitting into memory, using a proper DBMS is probably a good way out.
      --
      print map{chr}unpack(q{A3}x24,q{074117115116032097110111116104101114032080101114108032104097099107101114})
        Yes, I may try this sometime. For now, however, the problem was solved by upgrading DB_File to the latest version.