Publishing model with same node and different nodes dependencies

Edit online

The fictitious company, The XYZ Publishing, follows a business continuity model that involves separating the different platforms used to develop the web content. XYZ Publishinguses location dependency policies to keep some resource groups strictly on separate nodes and others together on the same node.

The Production database (PDB) and Production application (PApp) are hosted on the same node to facilitate maintenance (and perhaps the highest priority node for these resource groups has the most memory or faster processor). It also makes sense to set up a parent/child relation between them, since the application depends on the database. The database must be online for the application to function. The same conditions are true for the System Database (SDB) and the System application (Sapp) and for the QA Database (QADB) and the QA application (QAapp).

Since keeping the production database and application running is the highest priority, it makes sense to configure the cluster so that the three database resource groups stay on different nodes (make them an Online On Different Nodes dependency set), and assign the PDB resource group with the high priority. The SDB is the itermediate priority and the QADB is the low priority.

The databases and their related applications are each configured to belong to an Online On Same Node dependency set.

PowerHA® SystemMirror® handles these groups somewhat differently depending on how you configure startup, fallover, and fallback policies. It makes sense to have the participating node lists differ for each database and application set to facilitate keeping these resource groups on the preferred nodes.

The following figure shows the basic configuration of the three nodes and six resource groups.

Figure 1. Publishing model with parent dependencies or child dependencies and location dependencies

Resource group policies

For the sake of illustration of this case, all six resource groups might have the following behavioral policies:

Startup Policy: Online On First Available Node
Fallover Policy: Fallover to Next Priority Node
Fallback Policy: Never Fallback


Participating Nodes	Location Dependency	Parent/Child Dependency
PApp: 1, 2, 3 PDB: 1, 2, 3 SApp: 2, 3 SDB: 2, 3 QAApp: 3 QADB: 3	Online On The Same Node Dependent Groups: PApp with PDB SApp with SDB QAApp with QADB Online On Different Nodes Dependent set: [PDB SDB QADB] Priority: PDB > SDB > QADB	PApp (child) depends on PDB (parent) SApp (child) depends on SDB (parent) QAApp (child) depends on QADB (parent)