Abstracts of Magazine FUJITSU 2005-5 (VOL.56, NO.3)

Special Issue : Mission-Critical IA Server/Linux

• Trends in Open Systems and Related Fujitsu Activity

In April 2005, Fujitsu announced the release of its PRIMEQUEST of new IA servers. These servers are best-fit solutions for mission-critical applications, and one of their important characteristics is their thorough "openness." This paper first investigates the birth of open systems and the benefits they are having on users. It then looks at the rising popularity of open source software, particularly Linux. Lastly, this paper describes the hardware, OS, and middleware of the PRIMEQUEST from the viewpoint of open systems, focusing on the reasons Fujitsu is pro-active in the open-systems arena and open source development.

• Value Proposition of Newly Developed Mission-Critical IA Server: PRIMEQUEST

As today's broadband Internet society continues to evolve, market expectations for server platforms have changed. Aside from typical requirements for high reliability and high performance, customers also expect actual reductions in TCO (Total Cost of Ownership) and maximized ROI (Return On Investment). These end-customer benefits can be achieved through 24/7 non-stop operation, shorter recovery time, flexible system integration, improved efficiency by using open architecture products (including middleware and application software), as well as easy migration and improved maintainability and operability. Given its vast experience and know-how acquired over many years of developing mainframe and UNIX systems based on Linux/Windows and open standard IA technology, Fujitsu (in conjunction with major global players) stands in the forefront of developing mission-critical IA servers capable of meeting market expectations. This paper describes the value propositions of the newly developed PRIMEQUEST mission-critical IA server, its features, technologies, and target markets.

• High-Reliability Technology of Mission-Critical IA Server: PRIMEQUEST

The PRIMEQUEST of mission-critical IA servers are equipped with various RAS (Reliability, Availability, and Serviceability) functions from the ASIC (Application Specific Integrated Circuit) levels, unit levels, and up to the system level. These RAS functions provide the high reliability and high availability required for mission-critical operations on these servers. This paper first gives an overview of the PRIMEQUEST. It then discusses the speed-enhancement technology, scalability, high-reliability technology, and high-availability technology of the PRIMEQUEST, focusing on the high-reliability technology implemented in the system hierarchies.

• Fujitsu's Chipset Development for High-Performance and High-Reliability Mission-Critical IA Servers

Fujitsu has developed a new mission-critical IA-server named PRIMEQUEST in close collaboration with Intel using Intel's latest high-performance and highly reliable CPU (Itanium2, the 64-bit microprocessor). PRIMEQUEST offers linear scalability from a single CPU to 32 CPUs (with 64 CPUs in the second generation) and represents a highly reliable technology equivalent to that of a mainframe. In conjunction with this development, we also developed six new chipsets using cutting-edge CS101 ASIC technology and examined and successfully employed new technologies such as high-speed interconnection between chipsets, address and system mirroring (including inside the chipsets), and a new standard high-speed IO interface (PCI-Express). This paper describes an overview of the newly developed chipsets.

• Highly Reliable System Mirror Function of Mission-Critical IA Server: PRIMEQUEST

The system mirror function is an option for the PRIMEQUEST of mission-critical IA servers that dualizes the hardware components and operates them in clock synchronization basis. When this function is used and a failure occurs in one side of a dualized hardware component, the function continues processing in the other side. This function, therefore, considerably enhances system availability without the need for software intervention. Therefore it is implemented only via the servers' hardware, highly reliable and available system operation can be achieved without any modifications to the software available in a market.
This paper gives an overview of the system mirror function and discusses the benefits that it provides.

• Flexible IO Improving Flexibility and Reliability of Mission-Critical IA Server "PRIMEQUEST"

The PRIMEQUEST of mission-critical IA servers consists of the system board, which mounts the CPU and memory, and the IO unit (IOU), which mounts IO-related blocks such as hard disk drives and PCI slots. The system boards and IOUs are physically independent of each other and are interconnected by a crossbar. This feature is referred to as the "flexible IO" because it enables partitions to be configured by combining any system boards with any IOUs. This paper gives an overview of the flexible IO and discusses some of the benefits that it provides.

• Operation Management of Mission-Critical IA Server "PRIMEQUEST" for TCO reduction

System operation management functions are essential for stable operation of server systems; however, server system operation management at the moment is expensive. The PRIMEQUEST 400 series of mission-critical IA servers uses a server management processor (MMB: Management Board) mounted in the server cabinet to implement the functions required for operation management and enables all server management operations to be centrally controlled by the MMB's Web-UI function. The general-purpose LAN connected in the cabinet eliminates the need to change the LAN cable connection during system installation and configuration changes. These features significantly reduce the cost of server management. This paper discusses the various functions and characteristics of operation management that support stable operation of the PRIMEQUEST 400 series.

• Rapid improvements of LinuxOS ―Features and Reliability―

Rapid improvements continue to be made in the ongoing development and usage of Linux through a mailing list of volunteers known as the Linux community. These volunteers contribute source code, report information on any trouble to the community, and quickly provide required bug fixes. The latest kernel version (2.6) reflects the many improvements made to Linux, primarily in memory management and the process scheduler. Consequently, Linux has become a much more advanced OS. This paper describes the four features being developed by Fujitsu (in conjunction with the Linux community) for mission-critical systems. These include the 1) reliable crash dump feature (diskdump) used in case of a Kernel crash, for example, 2) enhanced MCA to minimize the effects of hardware failure and recover from failure by using MCA feature, 3) persistent device naming feature (udev) that prevents device names from being changed before and after maintenance and expansion, and 4) dynamic hardware reconfiguration (hotplug) for hot system maintenance and changing CPU, memory, and IO bus capacity.

• Linkage of PRIMECLUSTER and Mission-Critical IA Server "PRIMEQUEST"

Information Technology (IT) systems for today's ubiquitous age must address the urgent feature requirements for flexibly accommodating changes in large-scale workload and for 24/7 continuous operation. The "PRIMEQUEST" mission-critical IA server was developed as a core server in line with "open," "mission-critical," and "global" as the keywords. PRIMECLUSTER is basic software designed to maximize continuous operation time by enhancing availability through a redundant server, storage, and network configuration. PRIMECLUSTER has extended its competitiveness by incorporating a combination of Fujitsu's advanced middleware, servers, and storage. By combining Fujitsu's vast experience in high-reliability technology in collaboration with PRIMEQUEST and PRIMECLUSTER, we can provide cluster systems offering the highest reliability and longest continuous operation time in an open system. This paper introduces the technology employed and approach taken to achieve maximum availability through the linkage of PRIMEQUEST and PRIMECLUSTER.

• Application of Middleware Products to Mission-Critical IA Servers

Today's information technology (IT) systems need to operate 24/7, and more and more businesses require sophisticated, open-architecture IT systems that can operate in a network environment. However, as businesses acquire such systems, they also have to spend more time and money to keep them secure and reliable. In response, Fujitsu develops and provides highly secure and highly reliable open-architecture middleware products for mission-critical IA servers. These products are as reliable and safety as mainframe products. This paper describes Systemwalker, Interstage Application Server, Interstage Business Application Server, and Symfoware Server as these products and their features.

• LinuxOS Support in Fujitsu

With the appearance of commercial distributions of the LinuxOS to companies and the increasing number of available Linux applications, Linux is steadily being introduced in companies and local governments. Conventionally, Linux has only been used for the hardware and software of front-end systems, for example, Web servers and E-mail servers. Recently, however, Linux is also being used for middleware applications in mission-critical systems of companies and governments. These systems are required to operate 24/7 year-round, which makes it very important to provide Linux support services. This paper describes the Linux support provided by Fujitsu. First, it describes Fujitsu's cooperation with Linux distributors and the structure and policies of the support. Then, it describes the type of Linux support that Fujitsu provides and the support period.

• Applying Linux to Mission-Critical Financial Systems

The current trend of integrating financial services and developing market-based financial systems requires a transformation of business models in the financial industry. For the future computer systems of financial institutions, the need for "easy connection with an institution's own system or the systems of other companies," "shorter development periods," and "reduced total cost of ownership (TCO)" will become essential, in addition to typical requirements for high-performance, high-quality, and highly reliable business and back-office operations. Moreover, it is imperative to restructure existing complex systems. To resolve these issues, Fujitsu has provided a next-generation financial solution framework that enables large-scale mission-critical systems to be built on open systems environments such as Linux. This paper outlines and describes the characteristics of this solution framework.

• E-autonomy Solutions on Linux

The e-Japan Strategy, which aims to make Japan one of the most advanced IT nations in the world by 2005, has shifted from the infrastructure-development phase to e-Japan Strategy II, which will put the developed infrastructure to practical use. In the coming e-autonomy era, administrative services will respond to higher user needs and be more streamlined and less expensive to run. Realizing e-autonomy will require highly efficient, cheaper, and safer systems that protect society from hacking, computer viruses, and leaks of personal information to the Internet. Fujitsu will provide the best solutions for the problems of e-autonomy based on a Linux-based infrastructure. This paper introduces some of Fujitsu's Linux solutions for e-autonomy and examples of our sales activities in this field.

• Portal Construction Package for Universities: Campusmate/Portal

With the ever-growing spread of the Internet, people can now obtain information and services anytime and anywhere. Even universities and colleges are striving to provide services that improve the lives of their students by using a network environment. These days, close attention is given to constructing portal systems that provide students with personalized information on the campus. To help in this area, Fujitsu has developed and released a portal construction package for universities called Campusmate/Portal. This product operates on an IA (Intel Architecture) server running Linux, so it is easy to install and has excellent cost performance and security. It has already made the lives of students more convenient at several universities, where it has been installed and tested on the key IA servers. This paper describes how Campusmate/Portal is used on these IA servers and describes Campusmate/Portal's functions.

• Fujitsu's Work on Linux PC Clusters for High-Performance Computing

Recently, there have been significant improvements in the CPU performance and installed memory size of PCs. Also, open-source software such as Linux has recently been enhanced. Because of these developments, PC clusters are now being applied in high-performance computing (HPC). PC clusters were initially used as inexpensive, but troublesome, small-scale private systems and are currently used in many systems that are ranked in the TOP500 (www.top500.org). Similarly to off-the-shelf supercomputers, they have been used as main systems in large-scale computing centers. This paper outlines the trends of PC clusters and discusses some of the issues regarding their use as the main system of a computing center. It also describes Fujitsu's efforts to provide PC clusters to customers in the HPC area.