G
George Potemkin
Guest
> Read through data created by dump&load was approximately 15 times faster than read through data created afterwards, during normal production. Most likely it's caused by record fragmentation. Just recently I wrote a program that can analysis the record fragmentation in the specified /part/ of area as well as to do online defragmentation of the records (like idxcompact for indexes). Let me know if you would like to give it a try. Regards, George
Continue reading...
Continue reading...