Loading ...
Sorry, an error occurred while loading the content.

RE: [XP] When should you do performance optimization?

Expand Messages
  • Kent Beck
    Dear David, Some performance improvements simplify and clarify the design. Those should be done as soon as possible. Some performance problems need to be
    Message 1 of 10 , Jan 28, 2008
    • 0 Attachment
      Dear David,

      Some performance improvements simplify and clarify the design. Those should
      be done as soon as possible. Some performance problems need to be solved
      *now*. In either case, bending the design in the direction of performance is
      the right thing to do. The difficult situation is when you can see a way to
      complicate the design to meet future performance needs. Then you have
      certain cost now to potentially gain benefit and avoid cost later.

      My strategy is first to make sure I know where I stand with respect to
      performance. What I sketched as I was thinking about your question was a
      curve of the necessary performance over the next several years. I want the
      system to stay comfortably ahead of this curve.

      I have had a hard time writing deterministic performance tests. That's why I
      like quantitative measures (like the min/max/mean someone mentioned) and
      then tracking those over time. Even though they don't carry the reassurance
      of red/green, anyone who cares about performance can see where the system is
      with respect to performance goals and see trends so they can predict
      problems.

      Ken Auer and I wrote a paper called Lazy Optimization which summarizes my
      philosophy. I don't think it's available online, but the short form is to
      only optimize based on actual data and actual need.

      Regards,

      Kent Beck
      Three Rivers Institute

      _____

      From: extremeprogramming@yahoogroups.com
      [mailto:extremeprogramming@yahoogroups.com] On Behalf Of dpeterson72
      Sent: Monday, January 28, 2008 3:15 AM
      To: extremeprogramming@yahoogroups.com
      Subject: [XP] When should you do performance optimization?



      How much should you compromise today's simple design for tomorrow's
      potential scalability problems? How do you make the call?

      David
      ---
      David Peterson
      http://blog. <http://blog.davidpeterson.co.uk> davidpeterson.co.uk
      http://www.concordi <http://www.concordion.org> on.org






      [Non-text portions of this message have been removed]
    Your message has been successfully submitted and would be delivered to recipients shortly.