The software, called IBM Batch-on-Grid, automatically accommodates spikes and lulls in computing workload — the amount of work a system is handling at any given time — by allocating servers on the fly, helping ensure IT systems run around the clock. This allows organizations to continue operating during system failures, natural disasters or while complex applications are being updated across the Grid environment.
Batch-on-Grid blends self-managing autonomic computing with technology borrowed from a mainstay programming technique called “batch computing,” which originated in the 1950s and is still used by thousands of finance, government and industrial customers. In batch computing, various computing jobs are submitted to queues and then scheduled for processing. The new IBM software creates batch workloads within Grid computing environments, and then uses autonomic technology to automatically schedule and balance those workloads, which delivers capacity when needed and lowers costs.
Clients can automatically shift Grid workloads so that the most important jobs — those affecting customers and supply chain partners — are taken care of first. For example, an organization can limit a server to 95 percent capacity, and when a server approaches that threshold, the software will instantly shift the workload on demand to another server. This ensures customers and partners are provided with continuous service.
The software also allows IT departments to accurately forecast workloads for high-priority projects, such as getting new employees up and running after an acquisition, and ensures that enough capacity is available. In this way, companies can plan for computing resources to deal with high priority jobs and be able to budget and account for costs associated with those resources.
For example, a multi-national bank preparing to roll out a new online banking application across a Grid computing environment can predict the IT resources and capacity needed to support a customer service application.
Done manually, this is a complex process performed at multiple data centers that can take days, weeks or even months in the worst cases. By automating this process, the IBM solution cuts it down to a matter of minutes, while existing customer-facing applications continue uninterrupted. Once the application is deployed, the IBM solution ensures it will remain up and running by allocating the needed server resources according to changing workload requirements. It also provides the bank with a single point of control from which to manage the application across the Grid environment.
“As Grid computing grows in popularity, IBM is helping organizations simplify the management of grids and better respond to changes in their business and in the market,” said Bob Madey, vice president of strategy for IBM Tivoli software. “By incorporating self-managing autonomic software into workload management, we're able to help ensure our clients' systems remain up and running at all times, so they can do a better job of providing a high quality of service to their customers.”
The software can operate within a service-oriented architecture (SOA), so it can balance computing workloads in the most complex, heterogeneous IT environments — for virtually any type of application or computing job.
This prevents companies from having to spend time and resources developing multiple workload management processes for every different kind of application. As IT organizations are increasingly building SOAs to more easily integrate new and existing software on multiple platforms, the software allows them to manage their IT workloads regardless of the type of application or underlying technology is involved.
IBM Batch-on-Grid adds workload and capacity management components to IBM's IT Service Management offerings, IBM's set of software and services to help customers automate and standardize the way they design and integrate IT processes.
The solution manages batch workloads across Grid, mainframe, and distributed computing environments and consists of new versions of IBM Tivoli Workload Scheduler 8.3, which provides a single point of control for managing batch workloads in mainframe and distributed environments; IBM Tivoli Workload Scheduler Load Leveler, a job management system that allows customers to optimize workload performance on AIX and Linux systems; and IBM Tivoli Intelligent Orchestrator, which delivers computing capacity when needed by predicting impending workload requirements and provisioning the appropriate IT resources, such as additional servers.