Re: Page Cleaners/Idle Writes

In <24r2msINNf0k@emory.mathcs.emory.edu> mikea@hpdmmfi.boi.hp.com (Mike Alder) writes: >Dave, > Thanx for the response. I couldn't reach you directly. > page cleaners = LRUS is worth trying. > I remember seeing your post a couple of weeks ago when we were > discussing maximum performance - it was very imformative and it > was that reason I was experimenting with Page Cleaners. > *Do you understand how an Idle Write is tabulated. Why do I > continally see chunk writes when my pages are being flushed > outside of checkpoints. An Idle Write is defined as a write which is performed by a page cleaner which commenced activity as a result of its sleep period expiring (as opposed to an LRU write which is when a page cleaner is woken by a signal from tbinit). It is difficult to understand why you are only seeing chunk writes recorded if you are sure that you are getting page cleaner activity. I have seen a tbstat -F output on Pyramid which recorded everything as Async writes, with zeroes in the Fg, LRU, Idle and Chunk columns. This "feature" later disappeared from the same tbconfig, without any corrective action having been taken! (4.10.UF3 I believe) Does the HP 9000 (which I assume you are using) have a similar Asynchronous I/O extension? graeme -- Disclaimer: The author's opinions are his own, and not necessarily those of Pyramid Technology Ltd. or Pyramid Technology Inc. --------------------------------------------------------------------------- -m------- Graeme Sargent Voice: +44 (0)252 373035 ---mmm----- Senior Database Consultant Fax : +44 (0)252 373135 -----mmmmm--- Pyramid Technology Ltd. Telex: Tell who??? -------mmmmmmm- Farnborough, Hants GU14 7PL Email: graeme@pyra.co.uk --------------------------------------------------------------------------- We have the technology. The tricky bit is learning how to use it.