IBM Books

Administration Guide


Understanding system partitioning

System partitioning is a method for organizing an SP system into non-overlapping groups of nodes that can be dedicated to specific purposes. A group of such nodes (not including the control workstation) is a system partition. For example, you might want to configure a system partition to use for non-disruptive testing when you migrate to a new release of software. Or you might want to configure system partitions for multiple isolated production environments so that the workload in one environment will not interfere with the workload in another environment. Isolation of production environments is achieved by partitioning the switch in such a way that communication paths for different system partitions do not cross each other.

SP-attached servers and dependent nodes can be part of a system partition along with standard SP nodes. Partitioning is performed according to the switch port to which the server or dependent node is connected.

System partitions appear to most subsystems and for most user tasks as logical SP systems. It is important to understand that from an administrative point of view, each system partition is a logical SP system within one administrative domain.

On the control workstation, the administrator is in an environment for one system partition at a time. The SP_NAME environment variable identifies the system partition to subsystems. If this environment variable is not set, the system partition is defined by the primary: stanza in the /etc/SDR_dest_info file. Most tasks performed on the control workstation that get information from the SDR will get the information for that particular system partition.

In managing multiple system partitions, it is helpful to open a window for each system partition. You can set and export the SP_NAME environment variable in each window and set up the window title bar or shell prompt with the system partition name. The following script is an example:

sysparenv:
# !/bin/ksh
  for i in 'splst_syspars'
  do
     syspar='host $i | cut -f 1 -d"."'
     echo "Opening the $syspar partition environment"
     sleep 2
     export SP_NAME=$syspar
     aixterm -T "Work Environment for CWS 'hostname -s' - View: $syspar" -ls -sb &
  done
  exit
 
.profile addition:
# Added for syspar environment setup
  if [ "'env | grep SP_NAME | cut -d= -f1'" = SP_NAME ]
     then
        PS1="[`hostname -s`<p>$SP_NAME] ["'$PWD]> '
     else
        PS1="[`hostname -s`]["'$PWD]>'
  fi
  export ENV

You can check which system partition you are in with the command:

spget_syspar -n

You can achieve system partitioning either by applying one of a fixed set of supplied configurations or by creating a different system partition configuration and applying it. (See the book IBM RS/6000 SP: Planning Volume 2, Control Workstation and Software Environment for specific information on supplied configurations). The ssp.top option in the pssp install package contains a directory structure with all relevant files for partitioning the system with supplied system partition configurations. In addition, the ssp.top option contains the System Partitioning Aid, a tool for creating customized system partitions to suit your needs. Start the System Partitioning Aid from the Perspectives Launch Pad or with the spsyspar command. This option must be installed on the control workstation if you are partitioning your system into more than one system partition.


[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]