PRIMEQUEST

Performance, Reliability, and Flexibility in Open Systems

In 2005 when Fujitsu introduced PRIMEQUEST, they not only extended the boundaries of total reliability into the open systems world, but also laid the foundation for the most cost effective computing performance using the more popular environments of Windows and Linux. Barriers to the use of commodity products for mission critical operation were removed and the vast world of open applications became available for use in the most important and demanding business tasks. The cost benefits of open systems have long been known, but until PRIMEQUEST, many organisations seeking mission-critical operation had to forego those benefits because of reliability and scalability concerns. Now, with the introduction of new PRIMEQUEST 500 series, any lingeriing doubts about open systems performance v proprietry performance have been removed. The introduction of Dual-core Intel® Itanium® 2 processors (codenamed “Montecito”) shows the unlimited 64bit performance future of the Intel roadmap and the synergy between Fujitsu’s TRIOLE (Virtualization, Automation and Integration) strategy and Intel’s persuit of open systems performance.

PRIMEQUEST supports the rich capabilities of Windows and Linux systems in the mission critical arena with hardware platforms designed using mainframe-derived levels of hardware reliability, availability and serviceability (RAS), plus new levels of automation that remove the previous high costs of large systems administration. This leaves Windows and Linux customers free to enjoy investment protection, freedom of choice and unlimited growth when provisioning for future opportunites and business flexibility.

Further, PRIMEQUEST meets the needs of Open Systems organizations looking to maximise their existing Linux and/or Windows skills with top-to-bottom coverage in the data center. Growing organisations will no longer have to learn new skill-sets just to implement the highest levels of business support.

Designed using Intel® Itanium®2 processors with dual core configuration and a higher performance 533GHz front side bus, plus Fujitsu enhanced server technologies, garnered from decades of high reliability development; PRIMEQUEST conforms to Fujitsu's TRIOLE Open Systems' business continuity, business agility, and business efficiency standards for open systems. PRIMEQUEST, like all of Fujitsu's mission critical systems, also matches the expectations of customers looking at the next generation of "autonomous" computing and performance.

Business continuity comes from architecture with fault immunity included at every level. The Dual-core Intel® Itanium® 2 processors and Fujitsu chipset in combination, deliver the best industry standard implementation of RAS. At the system level there is unmatched error prevention and redundancy. The main components such as LSIs/circuits inside the chipset, system bus, I/O paths, in-built management console, and power supplies are duplexed. More over, the in-built "System Mirror" function enables superior reliability found in top-end mainframes. All addressing, data bus and memory operations can be duplexed for total operational redundancy. Errors can be isolated without system halt, for exemplary systems' operation continuity. Flexible I/O (FIO) means I/O resources can match real-time performance requirements on-demand. Such flexibility and standard redundancy means even maintenance downtime can be reduced by up to 90% compared with typical industry figures.

To further strengthen Linux for business critical use, about 500 Fujitsu developers are contributing to the open-source community. This and the work of the joint development organisation within Red Hat, is embodied in Red Hat Enterprise Linux AS (v.4 for Itanium) used with PRIMEQUEST. Fujitsu is also working closely with Microsoft, under the Global Alliance Partnership, to deliver fail-safe environments for mission-critical operation of Windows applications.

For business efficiency and flexibility new technologies allow up to 16 partitions (averaging 4 cores per partition) to operate safe from any single point of failure. For the first time, Fujitsu’s eXtended PARtitioning (XPAR) technology brings sub-systemboard hardware partitioning to open systems. The electrical isolation of each fully independent PRIMEQUEST partition provides the robustness and reliability characteristics of separate mission critical class systems, but with all the administration benefits of system consolidation and virtual resource allocation flexibility. Each Windows or Linux environment running in individual parallel partitions (including memory, CPU and I/O) can be independently and dynamically adjusted to match load requirements throughout the day. This means PRIMEQUEST is quick to adapt to change whilst always delivering excellent resource efficiency, flexibility and operational stability.

PRIMEQUEST's mission critical profile comes from full component redundancy and the System Mirror, where even part system boards are electrically isolated. Add in the separation of I/O from CPU and Memory, redundant high-speed cross bars and Fujitsu's PRIMECLUSTER environment, and it delivers the industry's fastest recovery times and up to 99.999% availability.

Typical of the attention to detail is the revolutionary cable management. Normally large servers require tens if not hundreds of cable connections. This management overhead is removed in PRIMEQUEST, as all cabling is integrated into the backplane. No physical cable adjustments are required; all connection changes are achieved via the management console using the functions of the MMB. This not only speeds reallocation of resources, it avoids cabling errors, and frees up precious PCI slots for other uses.

Cool running is also key to stable operation, as components last longer and work better within their normal heat tolerance range. In PRIMEQUEST Intel Itanium 2 processors use low voltage technology to enable greater scalability and improved performance in a smaller footprint. This is matched by Fujitsu’s efficient 90nm wiring technology and airflow design that reduces the overall power consumption by up to 50% that of other comparable sized IA architecture servers. . PRIMEQUEST has a refined server architecture based on Fujitsu LSI, chipset and system interconnect technology. This ensures it provides the best performance, smallest footprint and lowest power consumption in its class.

Partitioning of resources, in particular, is very robust. Fujitsu has chosen a granular hardware partitioning technique that removes any potential for total system failure. Not only are partition faults isolated from adjoining partitions, but the removal of partition control software removes any potential for that to become a single point of failure. On top of this partitioning, newly added eXtended PARtitioning (XPAR) function will double the partition granularity, which means twice the number of partitions are available compared to previous PRIMEQUEST models. Already the PRIMEQUEST fast restart capability greatly improves service recovery and reorganisation. Supported by the MMB's ability to pre-emptively detect faults, it enables the use of reserve system boards to configure and dynamically reallocate to problem partitions without a system halt. Into the future, Fujitsu is working with RedHat, Novell and the open-source community to add additional mission critical features to the virtual machine environment XEN. This will further extend the partitioning features of PRIMEQUEST and enable system virtualization that provides mission critical class dynamic reallocation of resouces between multiple operating environments.

Structured design and flexible interconnects ensure that only minimum resources are needed for each application. Systems can start small and grow as required. Processors, memory and I/O can be independently added using separate, flexibly configured, system boards, I/O Units and PCI boxes.

PRIMEQUEST reduces operational costs

PRIMEQUEST's flexibility makes it ideal for mixed workloads. Resource priority control enables the right processor resources to be set up for each task. Once minimum resource allocations are set any unused resources can be assigned to higher priority tasks as workloads and performance demands change. This reduces setup costs and enables more accurate and incremental addition of resources as demand grows.

For example, PRIMEQUEST is able to ensure that every application is a perfect fit to the available resources. Compare this with the use of typical IA servers. No resources are wasted and no applications need be constrained by fixed resource boundaries. PRIMEQUEST's 'on-demand' environment fully maximises return on infrastructure investment across the lifetime of the system, balanced with the best possible response to users at all times.

PRIMEQUEST reduces the cost of physical infrastructure management by allowing up to 16 concurrent Linux and Windows system partitions to operate in the same highly available system 'frame'. The in-built systems management reduces administration overheads and the reduced footprint of a single system saves floor space, reduces power consumption and centralises operational tasks.

The flexible I/O capability of PRIMEQUEST removes the cost of expensive add-on I/O port devices. I/O is always "independent" of partitions. Reconfiguration or partition swaps can be performed without cable changes and operational system restarts are achieved in minutes rather than hours.

By following the Intel 64-bit processor roadmap and enabling multi-generation processor upgrades, PRIMEQUEST extends the life of the overall IT investment. This means on-going and lower cost performance growth within the same environment and an extend life-cycle for the refurbishment and investment maximisation of major business applications.

Finally to ensure overall success of PRIMEQUEST, Fujitsu has entered into a large range of partnerships with organisations dedicated to development in the Linux and Windows space. This will ensure that PRIMEQUEST product development and innovation continues to match major developments and business trends as well as meet the requirements of users along the road to autonomous computing and beyond. Furthermore, Fujitsu is working with the Itanium Solutions Alliance (ISA) as a founding member for the expansion of an eco-system for the Lunux/Windows enterprise.

Major partners (in alphabetical order) include: BEA Systems, Inc., BMC Software, Inc., Computer Associates International, Inc., EMC Corporation, Intel Corporation, Microsoft Corporation, Network Appliance, Inc., Novell, Inc., Oracle Corporation, Red Hat, Inc., SAP AG, Softek Storage Solutions Corporation, VERITAS Software Corporation.

Fujitsu Systems Business (Thailand) Ltd.
Tel: +66 (0)-2302-1500
Fax: +66 (0)-2302-1555