Select a topic or year
BEAVERTON, Ore - 07 Dec 1999: -- IBM today announced two industry leading UNIX operating system enhancements for its NUMA-Q server line including improved workload management and integral support for next generation Storage Area Network (SAN) technologies.
The new features in NUMA-Q's DYNIX/ptx UNIX operating system, a member of the Project Monterey product line for Intel's IA-32 architecture, significantly increase the manageability, performance and availability in e-business and business intelligence environments. The enhancement of DYNIX/ptx is a step in the evolution of the Project Monterey product line. Future releases of the Project Monterey product, code named Monterey/64, for the Intel IA-64 architecture will incorporate enterprise features now available in DYNIX/ptx.
Dynamic Workload Management
DYNIX/ptx 4.5 includes a sophisticated workload management tool, Application Region Manager (ARM) with expanded capabilities allowing administrators to prepare and compensate for the radically shifting resource requirements associated with web-based applications.
Application Region Manager enables system resources to be dynamically allocated under control of the operating system and applied to specific workloads. Now in its fourth generation, ARM features a new easy to use graphical user interface, extensive monitoring and control per region, and ability to shift or modify resources without application termination, system reboot or user intervention.
"e-business and operational business intelligence applications are especially vulnerable to performance degradation and downtime caused by unpredictable loads applied to static resources. ARM places the customer in control by allowing the dynamic allocation of resources in a high-performance environment without bringing the system down," said Ralph Hedberg, Vice President of NUMA-Q Product Marketing at IBM.
In this release, ARM capabilities are expanded beyond processor allocation to include control of memory and kernel tables. Management of these resources enables application fault isolation, which can be used to reign in resource hungry applications. This is especially useful, for example, in business intelligence environments to help automate the isolation of rogue queries that can cause spikes in activity and cripple an entire system. ARM also facilitates server consolidation, within a single instance of the operating system, allowing IT organizations to deploy several applications or instances of an application without risk or degradation of application response time. Administrators are able to allocate system resources by department for more accurate cost center management.
First with Next Generation Fibre Channel Support
Also new with DYNIX/ptx 4.5 is integral support for Fibre Channel Class 2 and SAN diagnostics. The IBM NUMA-Q brand is first to offer operating system support for this new standard, which simplifies management and provide faster error detection and correction for SANs. These enhancements deliver significant performance gains for diagnosing both small and large-scale SANs.
Fibre Channel Class 2 increases the performance of SAN diagnostics by detecting and providing confirmation of delivery within milliseconds, along with a reason code for non-delivery. Detection by existing Class 3 standards takes up to tens of seconds or longer providing no explanation for errors. IBM NUMA-Q's Class 2 support enables NUMA Q's multi-path I/O subsystem to detect errors, and execute re-tries across a number of alternate paths. This helps ensure successful I/O and sustained application availability.
"NUMA-Q offers remarkable SAN and I/O subsystem performance. We believe that massively scalable I/O performance is the gating factor to successful e-business and business intelligence systems," said Hedberg. "Our commitment to the development of these technologies is reflected in our significant lead in large-scale, switched Fibre Channel SAN implementation and now in our first-to-market status with Class 2 support."
NUMA-Q SAN Diagnostics continuously collect diagnostic data across every component of the Fibre subsystem. Fully automated as a default capability, SAN Diagnostics helps reduce trouble-shooting time and provides an early warning detection by sampling all components of the SAN to capture the complete context of errors as they occur.
Other new features in DYNIX/ptx include cluster wide device naming, which simplifies administration of a cluster by providing a single system view of the sharable device configuration on all notes of the cluster, and Fibre Channel name service, which helps reduce the time a system spends in discovering and maintaining the fabric for increased SAN availability.