K
kunal.watkar
Guest
Hi, I have a task to dump and load very large database ( approx. 300GB) and I will be having not more that 32 hrs to complete this task. With conventional ASCII or Binary D/L this task is not achievable within given time limit. So, I tried parallel Dump/Load option. I used a raw-transfer to do parallel D/L . I wrote a 4GL code to transfer data of 1 table (tr_hist , size 55GB, records more than 120 million). After almost 6 hours of data transfer, only 30 million records was loaded into target database, which implies that for a complete table it will take around 24 hours. This is not suitable for my task's time limits. Please let me know which parameters I should tune to improve the performance. Please find below some details about my environment: Source DB: OS: AIX OE Version: 10.1A -L :500000 -bibufs :120 -s :81 -B :100000 -spin :40000 -aibufs :130 -directio Target DB: OS: Linux OE Version: 11.2.1 -L 300000 -bibufs 120 -s 81 -B 500000 -spin 1000 -aibufs 130 For doing raw-transfer, I am connecting source DB using -S parameter and progress 11.2.1 from Linux server.
Continue reading...
Continue reading...