Orchestrate and Simplify Workloads for IBM System z

Manage workloads better and get a better business outcome

Tivoli Beat - A weekly IBM service management perspective. For today's enterprise-class organizations, workload management can be a complex matter.

IT administrators find themselves asking questions such as: How do different systems interact at different points in the job stream? What roles do the mainframe and the distributed systems play in managing workloads? Should those roles change? What actions can I take today to make the infrastructure more agile, resilient and cost-effective in the future, especially given that I don't yet know how workloads will change?

One of the best ways to answer these questions is to simplify and consolidate overall workload management. When managers can easily establish and modify the way workloads are fulfilled, using centralized tools specifically designed for that task, overall costs will fall, yet performance will climb.

Furthermore, making changes to accommodate evolving business strategies or demand levels will also become more straightforward—translating directly into improved business agility and flexibility.

IBM Tivoli Workload Scheduler for z/OS: Lightweight, z-centric management

“When managers can easily establish and modify the way workloads are fulfilled, using centralized tools specifically designed for that task, overall costs will fall, yet performance will climb. Furthermore, making changes to accommodate changing business strategies or demand levels will also become more straightforward—translating directly into improved business agility and flexibility.” IBM Tivoli Workload Scheduler for z/OS (TWS for z/OS) delivers in all these respects. For organizations that have already installed TWS for z/OS, and also have distributed infrastructure, the new update is a great way to get enhanced business value via improved and simplified workload management. By shifting the focus from technical details to business goals and strategies, TWS for z/OS frees administrators to focus on what matters—ensuring that workloads are completed faster, more reliably and at lower costs.

And in the latest version of the solution, that argument becomes stronger than ever as a result of powerful new features and functionality.

Leading the list of those new features is the innovative, z-centric approach to workload management. TWS for z/OS now utilizes a lightweight, end-to-end architecture of agents, deployed on distributed systems, which serve as a control layer. These agents are managed by the TWS Controller, running on the IBM System z mainframe. The TWS Controller, in turn, is managed by administrators using any standard Web browser.

This design represents a considerable improvement for TWS for z/OS, as, it is responsible for primary management functions, and its power is being extended to distributed systems running operating systems from Microsoft Windows to Linux to UNIX. Such a mainframe-centric approach is a natural fit for shops in which the mainframe has long been seen as the logical central point of command. Yet it also takes full advantage of the flexibility and application library of the distributed systems.

Furthermore, via TWS for z/OS, the total cost of workload management will typically fall. Largely, this is because ongoing maintenance is more straightforward; workload modifications will typically no longer require specific adjustments to distributed systems; UNIX System Services (USS) doesn't need to be tweaked or tailored to accommodate new workloads involving UNIX-based distributed systems; a dedicated end-to-end server is no longer needed - the System z already owned by the organization is now being used for that purpose; and the protocol used by the TWS Controller to interact with TWS agents is the familiar HTTP—not a complex and potentially problematic communication layer.

The result? Easier, faster and less expensive management of workload topologies of all kinds, at every stage: from deployment to configuration to management.

An intuitive GUI and enhanced step-level control of job execution

Designing, visualizing and modifying workloads also becomes a straightforward matter in TWS for z/OS thanks to its powerful graphic interface. Job stream objects are now represented visually, right in the manager's Web browser, for easy comprehension and adjustment of how workloads are being fulfilled.

Several different views address various aspects of workload management:

Graphical view. Here, the idea is to model job stream possibilities in the TWS database. Drag and drop functionality makes this an exceptionally intuitive process.
Job Stream view. This view is suited to managing job streams in the plan. Workload dependencies and task requirements can be edited as required.
Impact view. What will the technological and business impact be of new job streams? This view helps to answer that question, allowing managers to navigate through the stages of the stream and determine the root cause of potential problems.
Plan view. This is a graphical representation of any selected set of job streams in the plan, allowing managers to determine how they are (or are not) mutually interdependent.

TWS for z/OS also includes new step-level dependencies, increasing the flexibility of how jobs are scheduled by branching the job stream based on different conditions. Jobs can now initiate even when previous jobs in the stream haven't fully completed, or have completed but yielded unusual outcomes. This feature also translates into improved business resilience, because unexpected errors won't always halt a job stream if a workaround process is available.

End-to-end scheduling improves fault tolerance

One compelling benefit of the new agent-based design in TWS for z/OS lies in the way it can be utilized to orchestrate end-to-end scheduling of tasks across the complete infrastructure.

How does this take place? Begin with the fact that the TWS agents on distributed systems communicate directly with the TWS Controller running on the IBM System z mainframe. This means there is an ongoing flow of status information in both directions.

As updates occur on the distributed network, the agents deployed on those systems report to the TWS Controller (which serves as a master domain manager). TWS can then update the plan based on that new information, and then notify the distributed architecture of the new plan.

The upshot for workload managers is improved overall fault tolerance. Thanks to the end-to-end support included with TWS for z/OS, workloads can be orchestrated in such a way as to bypass technical failures in even the most complex IT infrastructures.

Workload service assurance: SLA fulfillment made simpler

How can organizations ensure that the most critical tasks are always executed—and on time?

A key strength of TWS for z/OS is workload service assurance. TWS gives workload managers a color-coded interface that reflects both crucial jobs and their performance/risk levels. If targets aren't being met (or aren't likely to be met in the near future), action can be taken to reduce or eliminate the problem, before it impacts the business.

That action, in fact, can often be taken by TWS itself. Thanks to workload service assurance, TWS is capable of first determining the path the workload takes through the infrastructure, then making dynamic adjustments to solve different types of problems. Often, for instance, it will do so by changing the processing priority of key tasks running on distributed systems. With the higher priority will come higher performance for that task, translating into the overall workload being completed within the intended time.

And for organizations whose workloads are characterized by service level agreements (SLAs) with mandatory terms, this can make a tremendous difference to the business bottom line.

Learn more

Contact IBM

Considering a purchase?

Or call us at:
877-426-3774
Priority code:
109HJ03W

Pulse 2010 Virtual Experience

Attend sessions, breakouts and demos and find Pulse assets full of hints and tips, user experiences and key technical content that you can't get anywhere else.