The implementation of these monitor buffers is inconsistent with the VM monitor architecture. They are NOT updated consistently based on time, but consistently based on CPU consumption. Thus you could NEVER correlate the Linux activity to the guest machine except by occasional accident. There are right ways to implement APPLDATA monitor records and wrong ways. Consistency seems to me a requirement but it can't be accomplished with this current implementation.
So in the words from a recent customer visit, what is better, no number, or bad numbers? good numbers might cost a bit more.... >From: Rob van der Heij <[EMAIL PROTECTED]> >Mike Fry wrote: > >>The data is then processed using MXG/SAS. Inspection of the data shows that >>the records (apart from the date/time fields) are all the same. >> >> >> >No doubt others who did play with this will know better, but >iirc this will happen on an idle server because the reading the >kernel meters and updating the contents of the buffer for >monitor data is not driven by wall clock time. The idea is that >the meters don't change much when there is no work done, so it >is driven off the virtual CPU timer instead. Do you also see >this on an active server? > >Rob > "If you can't measure it, I'm Just NOT interested!"(tm) /************************************************************/ Barton Robinson - CBW Internet: [EMAIL PROTECTED] Velocity Software, Inc Mailing Address: 196-D Castro Street P.O. Box 390640 Mountain View, CA 94041 Mountain View, CA 94039-0640 VM Performance Hotline: 650-964-8867 Fax: 650-964-9012 Web Page: WWW.VELOCITY-SOFTWARE.COM /************************************************************/ ---------------------------------------------------------------------- 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
