why we have Pools in config file?

rajasingam · Post by **rajasingam** » Tue Mar 04, 2008 11:36 pm

Hi friends,

What is the use of Pools in configuration file and how to assign it?

Thanks in advance..

Regards,
Rajasingam

tkbharani · Post by **tkbharani** » Wed Mar 05, 2008 1:05 am

In one line its Collection of nodes or pools(disk and scratch disk pool).
At certain times, we run some stages or some specific jobs which requires more physical storage space (staging area). At these times we need more nodes for processing. We can also call this as reserve storage area. Example of nodes configured in a pool
----------------------------------------
node "node1"
{
fastname "node1_css"
pools "" "node1" "node1_css" "pool4"
resource disk "/orch/s0" {}
resource scratchdisk "/scratch" {}
}
Once we have defined a node Pool , we can force a Parallel job to run on the specific pool and nodes,through which we get more space for processing.
For more information read Parallel Job developers Guide -Configuration Files(Pg 1015)

ray.wurlod · Post by **ray.wurlod** » Wed Mar 05, 2008 2:31 am

That's not quite correct. It is operators (stages) that can be restricted to run in pools, not jobs.

Pools are more useful in an MPP environment, where - for example - database client or other software might not be installed on all nodes. You can - in this example - create a pool containing only those nodes on which it is installed, and constraint the pertinent stage to operate only on those nodes.

For example you may have a 36-node configuration, but only have DB2 client software installed on 24 of those machines. You could create a DB2 node pool containing only those 24 nodes. The DB2 stage(s) in the job can then be constrained only to use those machines.