kjkoster
09-12-2008, 08:31
Dear All,
I'm at the Devoxx Java developer conference. While the whole conference is decidedly developer oriented there are Kirk Pepperdine's talks. They should interest Java developers and system administrators alike.
Sun was right to decide to retire jconsole in favour of VisualVM. It allows you to do so much better analysis that it's not even possible to compare the two. :-)
Kirk led a BOF session, where we did some live tuning of a small Java application. What was most interesting was not so much the rigid process that Kirk applies to tuning, but the response of the audience.
What I noticed was that the members of the audience lept to conclusions about what the problem was. Kirk had a hard time keeping everyone on the process path. Like a herd of cats, developers would leap to unfounded conclusions. Some of these conclusions were (accidently) right and some were wrong.
The lesson I take home is: keep going back to the evidence the actual benchmark provides. Did I leap to a conclusion, or does the evidence support my thesis?
Kees Jan
PS. Tonight I'll have my talk in room 6 from 17:25 to 17:55 (off the top of my head). See you there.
I'm at the Devoxx Java developer conference. While the whole conference is decidedly developer oriented there are Kirk Pepperdine's talks. They should interest Java developers and system administrators alike.
Sun was right to decide to retire jconsole in favour of VisualVM. It allows you to do so much better analysis that it's not even possible to compare the two. :-)
Kirk led a BOF session, where we did some live tuning of a small Java application. What was most interesting was not so much the rigid process that Kirk applies to tuning, but the response of the audience.
What I noticed was that the members of the audience lept to conclusions about what the problem was. Kirk had a hard time keeping everyone on the process path. Like a herd of cats, developers would leap to unfounded conclusions. Some of these conclusions were (accidently) right and some were wrong.
The lesson I take home is: keep going back to the evidence the actual benchmark provides. Did I leap to a conclusion, or does the evidence support my thesis?
Kees Jan
PS. Tonight I'll have my talk in room 6 from 17:25 to 17:55 (off the top of my head). See you there.