Clarification required about Cluster Environment

maffan76 · Post by **maffan76** » Sun Oct 23, 2005 3:34 am

Hi,
I want to clarify that we have a windows cluster running; one machine conatins the OS and the other contains Data Drives, so from DS point of view is it a single node or a multinode system? because incase of multinode system the installation of DS-PX will change interms of RSH config etc and will have to setup extra ENV variable. Or DS has it own definition of cluster e.g. N independent computers are connected by installing DS server on each machine and confguring RSH for comunication among the nodes.

Thanks in advance.

vmcburney · Post by **vmcburney** » Sun Oct 23, 2005 6:04 pm

It depends on what resources you are "borrowing" from those other machines. Assuming your DataStage server gets installed on the OS machine do you anticipate this machine has the CPUs and RAM required or will you be borrowing CPU and RAM from the cluster? Or are you only using them for disk space? Would be easy to link them to the OS server as network drives and reference them in the node configuration file.

maffan76 · Post by **maffan76** » Mon Oct 24, 2005 12:26 am

Thanks vincent,
But Whats the default config of DS and hows does it see the cluster during the installation as it does not ask anything. so to avoid any licnesing prolem i want to understand its behaviour.

Thanks in Advance.

ray.wurlod · Post by **ray.wurlod** » Mon Oct 24, 2005 12:43 am

The default configuration is the one you see as "default" in the configuration file viewer/editor in Manager client.

maffan76 · Post by **maffan76** » Mon Oct 24, 2005 12:51 am

So another question,
Can i change the paths of DataSet and Scratch for each node differently, i mean for node1 Drive C,node 2 Drive D and So on?

ray.wurlod · Post by **ray.wurlod** » Mon Oct 24, 2005 1:15 am

Yes.

Read the chapter on configuration files in the Manager manual, particularly the sections on resources.

Data Sets are automatically partitioned across all the processing nodes in the configuration, unless explicitly constrained into a node pool or a disk resource pool by means of stage properties.

maffan76 · Post by **maffan76** » Mon Oct 24, 2005 1:36 am

Thanks Ray, I'll read the chapter

vmcburney · Post by **vmcburney** » Mon Oct 24, 2005 5:46 pm

You should get better results if you can split your node resource and scratch disks onto different disk drives as this should ease I/O bottlenecks.