[Sorry, the "virus" has been removed. Nannyware c h o k e d on b i t c h e n. Still wondering what set it off for the Monday virus-l digest, among others.] Peter Ross Sharman said: >It's actually something I discussed with a publisher at OpenWorld and they >want me to throw together a plan on what would be in such a book and then >possibly write it. Any ideas you'd like me to include? ;) The various options, how they've evolved over time, what they are appropriate towards. Oracle v. others definitions of streams, clusters, and other buzzwords. Needs definition. Timeline of bug fixes, workarounds and bugs never owned up to. How things work[ed] v. how things were/are supposed to work. Performance implications. Example "normal" installations. How to make typically strange configurations work (ie, bunch of windows boxes with indeterminate networking and generic obscure crappy RAID-5). Super-b i t c h e n configurations. Automation of testing of various failure scenarios. If or Why Oracle is better than Rdb. Why or If RAC is more than just string and bubblegum over commodity hardware. Why or If "commodity" doesn't mean "crappy." Why average places need RAC. Answers to every HA question ever asked on oracle-l or cdo* or metalink. When to say "we need to update the SLA." Oh, and keep it under 1000 pages, please. ;) Joel Garry http://www.garry.to -- //www.freelists.org/webpage/oracle-l