Sun Cluster Data Services
By dlnprasad on Nov 14, 2006
Sun Cluster software provides the framework and API required to make applications highly available on Solaris OS. The software application, designed for solaris, does not have to be modified to use Sun Cluster API. Instead, you write an agent which acts as the interface between Sun Cluster core software and the application. Sun Cluster comes bundled with a rich portfolio of agents, also called data services, which make applications highly available on Sun Cluster software. These agents, designed and developed by the Sun Cluster Engineering team, have undergone rigourous testing by the internal Quality Assurance engineers.
Agents for the the following software products are available in Sun Cluster 3.1 and the upcoming Sun Cluster 3.2 releases. The following list is not in any particular order:
Oracle's Siebel CRM (server and gateway)
BEA WebLogic Server
NFS Server (SC 3.2 supports the latest NFS V4)
Apache Web Server
Sun Java ES Web Server
Sun Java ES Application Server
Sun Java ES Message Queue Broker
Oracle E-Business Suite
Oracle Application Server
IBM WebSphere MQ
Sun N1 Grid Engine
Sun N1 Service Provisioning system
SAP Web Application Server
MaxDB (previously called SAP DB)
In some cases I might not have listed the actual name of the product, as listed in the product manuals of the ISV. Please check with the ISV for the exact product name. The above list is provided to give you a high-level view of the rich application support on Sun Cluster software. If you want more details on how to configure and administer the agents for the above applications, please refer to the Data Service administration guides available at:
As you can see, the list is very long. In addition to the agents from Sun Cluster Engineering, some internal Sun product groups and ISVs have agents for Sun Cluster software. IBM has designed and developed agents for some components of the IBM Informix and IBM DB2 product family. Symantec supports a Sun Cluster agent for Veritas NetBackup. The Sun Java ES Directory Server, Sun Java ES Messaging Server, and Sun Java ES Calender Server groups have written agents for Sun Cluster software which are bundled with the respective products.
If your application is not in the above list, then there is nothing to worry about. It is very easy to write an agent for Sun Cluster software by using the Data Service Development tools available in the product.
You need to first check if your application can be made highly available on Sun Cluster software. This chapter in the Sun Cluster Data Services Developer's Guide lists everything you need to verify your application's cluster readiness. Most of the applications can be integrated with Sun Cluster software right out of the box. Sometimes you might have to enhance the application a little bit to be able to integrate with Sun Cluster software. Once the application is ready for integration, use the Sun Cluster Agent Builder to generate an agent for you. The Sun Cluster Agent Builder not only generates the code for you but also generates the Makefiles to compile the code and build a nice Solaris package. This generated package can be easily installed by using the pkgadd utility.
If you do not want to write a seperate agent for your application, you can use the Generic Data Service (GDS) agent to make your application HA on Sun Cluster software. Generic Data Service, as the name suggests, is a generic agent designed by Sun Cluster Engineering. GDS takes as an input, scripts to start, stop, validate, and probe an application. Instead of writing a separate agent for your application, you just write scripts to start, stop, validate, and probe your application and supply these scripts as extension properties to the GDS resource at the time of creation.