ORACLE12C one way to manage job resources

Database: 12.1.0.2,rac,cdb mode

The author is responsible for moving two 12.1.0.2 cdb clusters, one on AIX, and one on Linux, unfortunately, they are all mixed and the data is more than 100 T.

Since the other departments are delivered, already 12c, before 12c is not very familiar with, but still want to see whether it can be without the premise of the library to maximize performance.

The result is not, because some OLTP-biased applications tend to feel slow.

What to do? Had to specify the node through service, at least two nodes per service, the service internal use TAF configuration, a master node, a standby node.

For example, the services created for an OLTP are as follows:

Srvctl add service-db wgdb-service nsn_flowdb-preferred wgdb2-available wgdb4-tafpolicy preconnect-policy Automatic -failovertype session-failovermethod basic-failoverretry 180-failoverdelay 2-pdb nsn_flow

and on Node 2, it is only assigned to all of these OLTP type applications, other OLAP types on other nodes.

Personally think, on this point, Oracle database is far superior to all kinds of RDBMS and bigdata, light this aspect, can save a lot of enterprise resources and cost.

The resource management of other RDBMS is a bit complicated, far less convenient than Oracle.

In other words, Oracle is doing so conveniently to get closer to cloud.

After a while, most of the time, OLTP applications are fast, but occasionally there are slow situations, such as a certain point in the morning or afternoon.

What's going on? Then it analyzes the running log of the job, and discovers that the job itself does not run according to the created service, but on all nodes, it is distributed on each node according to the Oracle database's own equalization algorithm.

We also studied the work of the next Oracle and found that the service can be controlled by scheduling classes, and I'm glad that we don't need any complicated operations.

But here's the problem:

1. The existing job is using the default scheduling class, and the default scheduling class does not specify a node

2. Subsequent developers will use the default scheduling class if they are not forced

The decision to modify the default scheduling class for the other PDB:

beginDbms_scheduler.set_attribute (Name= 'Default_job_class', attribute= 'Comments', value= 'This are for all OLAP apps'); Dbms_scheduler.set_attribute (Name= 'Default_job_class', attribute= 'Service', value= 'Wybigdata'); End;

As for, the process of running OLAP on a particular node is slow and does not have any relationship, and OLAP can tolerate this situation.

Given that Oracle's schedule is so powerful, subsequent decisions are forcing vendors to use schedules instead of jobs, which are bad, and should have been thrown into the historical garbage heap.

In addition to controlling resource allocations through service, Oracle can also manage resource allocations for scheduled jobs in other ways, such as through resource groups.

Therefore, in the early stages of building a cluster, it is very important to choose the configuration of the cluster.

ORACLE12C one way to manage job resources