High Availability Scenarios With IBM Tivoli Workload Scheduler And IBM Tivoli Framework

In this appendix, we describe the implementation tasks within a deployment of a HACMP Version 4.5 and IBM Tivoli Workload Scheduler Version 8.1 scheduling environment in a real customer. We cover the installation roadmap and actual installation steps, and provide our observations in this real-life implementation.

The versions of software used are HACMP.

Rationale for IBM Tivoli Workload Scheduler and HACMP integration

The rationale for the integration of IBM Tivoli Workload Scheduler and HACMP was to use a proactive approach to a highly available scheduling solution, rather than a reactive approach. The IBM AIX/SP frame hardware environment has been an impressively stable system. However, on occasion as a TCP/IP network issue arises, customers new to IBM Tivoli Workload Scheduler scheduling environments naturally become concerned that IBM Tivoli Workload Scheduler schedules and jobs are not running on FTAs as expected. It is then realized that the IBM Tivoli Workload Scheduler FTAs continue to run their jobs even during these temporary network disruptions. This concern then developed into a risk assessment where the actual loss of the IBM Tivoli Workload Scheduling Master Domain Manager was considered.

Taking the loss of a IBM Tivoli Workload SchedulerMaster Domain Manager into consideration can be a serious concern for many customers. Where some customers feel a IBM Tivoli Workload Scheduler Backup Domain Manager is sufficient for a failover scenario, other customers will realize that their entire data center, which is now controlled by IBM Tivoli Workload Scheduler, could potentially go idle for several hours during this failover period.

This could be a very serious problem for a large customer environment, especially if a IBM Tivoli Workload Scheduler MDM failure were to occur shortly before the release of the (05:59) Jnextday job. Data centers running business critical applications or 10000 to 20000 jobs a day simply cannot afford a lapse in a scheduling service. Therefore, a highly available IBM Tivoli Workload Scheduling scheduling solution must be implemented.