C
ChUIMonster
Guest
Your checkpoints are pretty close together -- increasing the bi cluster size would be a good idea. You have waits on the OM latch. Increase -omsize to be at least as large as the number of tables & indexes in your db. There are a lot of LRU latch waits. You could implement -lruskips 100. ProTop is your friend. http://dbappraise.com If you seed it with an appropriate dbanalys/dbname.dba file then it might also tell you some interesting things about where the table and index hot spots are (-tablerangesize and -indexrangesie need to be large enough too) If that shows that you have some very small tables that are extremely active (which is surprisingly common) then you could potentially move those tables into -B2 and stop them from hitting disk all the time (this, of course, assumes that they are in type 2 storage -- even if they are not, if they are small then the effort to move them isn't hard and could be even more beneficial)
Continue reading...
Continue reading...