PPT Slide
What to do about BR, BTF, and RAU
Bufwaits Ratio (BR) - ideal ɟ% | 7-10% struggling | ᡂ% service death
- Increase buffers if BTF or cache ratios indicate this is a problem
- Increase LRUS (avoid 64, 72 & 96. CLEANERS >= LRUS)
- Alter LRU hashing policies with LRUPOLICY
- More CPU VPS may also help (keep restrictions on NUMCPUVPS in mind)
Buffer Turnover Frequency (BTF) - ᡂ mins (or BTR ɞ)
- Look for page flag bug (Rel ɟ.31UC6 or indexes created by earlier rel.)
- Shift load so queries on each server are more homogeneous
Readahead Utilization Ratio near 100%
- Reduce RA_PAGES and/or RA_THRESHOLD
- Faster more intelligent disk farm