David Kreuter wrote:
I vote no numbers. Bad numbers = bad decisions.
You can have them right now then. Happy? How about numbers that only reflect the load of computing those numbers?
I believe it depends on the type of numbers. If someone is asking me about utilization of the filesystem, I don't think it is necessarily bad to say "I checked 9:30 and it was 43% and there has not been happening a lot since then" I think it would be worse to say the usage is 43% at 9:35 when you know last time you checked was 9:30. If you look at something like RMF-PM, it is clear what price you pay to check this type of metrics every 30 seconds even when the system would have been idle. An idle server may use some 5 mS of CPU in a minute when idle. It would be a waste to go in and ask what processes it spend those 5 mS on (knowing that collecting that info will cost a multiple of those 5 mS).
Am I wrong that another popular performance product also has the feature to reduce the frequency of reading the meters when the Linux server is idle? Would that not also reduce the accuracy with which you report statistics?
Rob
---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
