For Oracle support & training call (800) 766-1884
Free Oracle Tips

Home
Oracle Tips
Oracle Code Depot
Oracle Monitoring
Oracle Consulting
Oracle Training
Oracle News
Oracle Forum
Oracle Support





 

Free Oracle Tips

image

 
HTML Text

Free Oracle App Server Tips

image

 
HTML Text


Privacy Policy

Redneck

Dress Code

Oracle tuning

Oracle training

Oracle support

Remote Oracle


 

 

 

 

Donald K. Burleson

Oracle9i RAC Tips

What is VSD?

The IBM VSD (Virtual Shared Disk) software is a component of PSSP that allows data on disks attached to some SP nodes to be shared with other nodes using a high-speed interconnect network like SP Switch or SP Switch2. The Parallel System Support Programs (PSSP) software provides a comprehensive suite of applications for the installation, operation, management, and administration of the RS/6000 SP attached servers and Clustered Enterprise Servers (CES) from a single point of control.

The VSD is a logical volume that can be accessed both by the node to which it belongs, and by other nodes in the system. The VSD nodes that share data may reside in a single system partition. VSD software allows applications running on different nodes to have access to raw logical volumes as if they were local. I/O requests to VSDs are routed by the VSD device driver, which is loaded as a kernel extension on each node, thus making raw logical volumes accessible to other nodes in the system.

Depending on the VSD function, a node can be:

  • A VSD server that has local attached disks. It is able to complete I/O requests from VSD clients by using a communication network inside the system.
     

  • A VSD client that is a node requesting access to VSDs.

There is another variation for VSD, which is known as Concurrent VSD. The Concurrent Virtual Shared Disk feature of PSSP allows multiple VSD servers to simultaneously access logical volumes inside a volume group using the Concurrent Logical Volume Manager (CLVM) component supplied by AIX. I/O requests from nodes that do not have locally attached disks are spread across VSD servers, thus improving raw logical volume access. When we use the Concurrent Virtual Shared Disk, recovery from node failure is much faster because the failed node is marked as unavailable to all other nodes. Its access to the physical disk is fenced, while the other nodes can continue to access the disks.

Virtual Shared Disks provide access to non-local disks. They behave like raw logical volumes. They can be configured using SMIT or commands. VSD(s) are accessed through the SP Switch so the network overhead is very low.

Configuring VSD for Oracle use

The VSD layer is over the AIX logical volume manager layer. When Oracle wants to access a file, it uses the VSD name of the file. VSD resides only with raw devices. The VSD layer sends the I/O request to the machine that can physically access the disk. In case of a problem on a machine, the VSD configurations file are updated and all the I/O previously processed by the broken machine are rerouted to the remaining node.

It is not a concurrent disk access, but a shared access. Only one node has access to a particular disk. All the I/O requests to a disk are transmitted to the node that owns the disk. This node handles the physical read/write and sends the data to the requester node.

For example, in a two-node cluster, at the LVM layer, there are two volume groups. Each volume group is vary-on and used on one node and vary-off (but defined) on the other node. The volume groups are not opened in a concurrent mode. They are shared between all the nodes but they are vary-on on a single node. In case of a hardware problem with a node, the other node of the cluster can vary-on the volume group of the broken machine and access both of the volume groups.

Oracle uses IBM VSD software to allow instances to access disks that are physically attached to other nodes. Data to be exchanged between nodes is passed through the IBM SP Switch. VSD file sets are installed on CWS and all nodes.

To enable the VSD of a SP, execute the following steps:

  1. Define the nodes of the cluster – the primary and secondary and the link between them.

  2. Define the VSD volume groups. For each group, indicate the node that normally accesses the volume group, and then define the backup node.

  3. Define logical volumes (raw devices) that can be used by VSD. A new VSD name is set which references the volume group and the logical name.

  4. Start the VSD.

For more details see the white paper ‘Quick Installation Guide ORAC on IBM pSeries with AIX’ – Fabienne Lepetit and Michel Passet.


For more information, see the book Oracle 11g Grid and Real Application Clusters  - 30% off if you buy it directly from Rampant TechPress .  Written by top Oracle experts, this RAC book has a complete online code depot with ready to use RAC scripts.


For more details and scripts, see my new book " Oracle Tuning: The Definitive Reference", over 900 pages of BC's favorite tuning tips & scripts. 

You can buy it direct from the publisher for 30%-off and get instant access to the code depot.

 

 


 

 

  
 

 
 
 
 
Oracle performance tuning software
 
 

 

 
 
 
Oracle performance Tuning 10g reference poster
 
 
 
Oracle training in Linux commands
 
Oracle training Excel
 
Oracle training & performance tuning books
 

Fast-Track Oracle Support
PO Box 511
Kittrell, NC 27544


Email BC: