Query related to RAC architecture...
2723794Oct 8 2010 — edited Oct 21 2010Hi All,
I have few question related to ORACLE 10g/11gRAC setup. I guess internal architecuture of both won't be change much except adding of new concept "Scan IP" and might be some other which I dont know perfreclty.
I have two node cluster setup.
My question are:
1- if I want to run batch job, which is actually performing DML statment [ combination of insert,update and delete] so It's recommended to run batch job in all nodes. Because at the same time other operation is also performed on this RAC setup and it might be slow down DB performance because bulky DML transaction. Is it advisable to run this batch on just single node?
2- if I want to run some jobs [dbms_jobs] on specific nodes then how can I control it?
3- if I want to perform single node action in RAC environment, then I can does this by giving "alter system set cluster_database=false scope=spfile sid=*" [ like enable archive log mode] . So what is the difference between run this ALTER command and if I down one node through "srvctl" command line utility?
4- In my application I have few transaction which is having nested loops and internally called procedures/triggers and all mix type of workload. If suddenly one node goes down then what happen if my transaction runninh on this nodes? I read few pdfs and came to knwo that all runnign transaction needs to be rollback so what happen if my application internally already commit at some level? does this measn inconsistent data?
IF anybody having any idea about any point out of these 4 points then guide me please.
Thanks...