NOTE: this is an archived page and the content is likely to be out of date.
In 1954, Fujitsu released FACOM100, the first relay computer to go on general release in Japan. Since then, it has been developing and providing new technology, under the theme of high performance and high reliability, to meet customer requirements amid the paradigm shift in ICT. Currently, the computer paradigm is changing owing to changes in corporate needs for ICT, technological innovation (e.g. virtualization technology), and changes in social environments such as global warming. There are needs for servers, which are part of ICT and Cloud computing infrastructure, to cope with these changes. This paper introduces Fujitsu's efforts for server strategies, server development and new server platforms against the background of the new computing paradigm and developments in Cloud computing.
SPARC64 is the name of the series of SPARC-V9 architecture processors that Fujitsu has developed. The development of the first SPARC64 started in the 1990s, and the development of the latest generation is actively on going as of 2010. The processor frequency of the first SPARC64 was as little as 118 MHz and it had only 20 million transistors, while the processor frequency of the latest SPARC64 VIIIfx is 2 GHz and it has more than 700 million transistors. The SPARC64 series has evolved to meet the needs of a new era. This paper describes the history of the SPARC64 processor development, and the enhanced points of the each generation.
This paper introduces technology used in designing high-performance processors installed in servers so that they can distribute high-frequency clock signals among the whole chip. Clock synchronization systems must be designed so that clock signals can reach all synchronous sequential circuits in the processor at the same time. The influence of interconnect inductance cannot be ignored because clock frequencies have been getting higher, chip sizes have been getting larger, and low-resistance interconnects have appeared. Therefore, it is necessary to give sufficient consideration to inductance when designing such systems. The technology introduced in this paper uses both circuit design techniques to suppress clock skew and crosstalk noise, and CAD techniques which employ these techniques in an actual chip. The influence of inductance in clock transmission can be analyzed electrically by simulating actually designed clock circuits at the design stage, and this can help lead to higher performance processors.
We are developing high-speed interconnect technology for servers to meet customers' needs for transmitting huge amounts of data as quickly as possible. High-speed interconnect technology consists of high-speed I/O circuits and high-speed printed circuit boards (PCBs), and these technologies are optimized to achieve the target performance. This paper explains the architecture of I/O circuits that are part of efforts to develop high-speed transmission circuits. It also introduces the results of testing a prototype chip that was built to evaluate 20 Gbps transmission. It goes on to explain efforts for improving simulation accuracy so that high-speed signals can be sent stably. For accurate simulations, we think it is important to accurately evaluate each part that makes up the transmission line and reflect the results in a simulation model.
Fujitsu is the only company in Japan that develops its own processors for use in server products that support the social infrastructure. Its processor development strategy is to collaborate with the internal semiconductor group and simultaneously develop the processor and semiconductor technology. This paper introduces SRAM development technology, which is a complex technology combining both semiconductor manufacturing and circuit systems. It fully meets conflicting server processor requirements such as high performance, small area and low power. It is a technology that is essential for starting up new technology and having it fully operational at the same time. Level-1 cache SRAM speed determines the processor clock rate, while the data processing bottle-neck is determined by the density of the level-2 cache SRAM. Thus SRAM is a key technology for server processors. As finer semiconductor technologies progress, various problems arise and the variability of the memory cell in the SRAM gets bigger. Consequently, development of SRAM that meets the server processor requirements is getting critical. This paper describes our SRAM development methodology.
There is an emerging trend for green computer servers. This trend includes the enhanced Top Runner Program in Japan, new requirements of EuP Lot 3 in the EU, and the ENERGY STAR program for computer servers in the US. These have focused customers' attention on reducing environmental hazards when using servers. Fujitsu has been working to develop servers that have a lower environmental load. It is doing this by introducing green IT technologies to reduce power consumption, save resources and improve the recycling rate. This paper describes specific examples of Fujitsu's efforts in the areas of power, cooling and mechanical technologies for green computer servers.
Server products are at the core of systems which assume critical missions in companies or society, and are required to operate stably. With the dramatic advances in technology, the environment surrounding server products has been changing. For example, there has been progress in virtualization or power saving in addition to high functionality and high performance, and even the adaptation of Cloud computing. Likewise, customer needs for quality also have been changing. Accompanying such changes, the quality assurance of server products has been reviewed to ensure stable operations by optimizing the development process or developing evaluation techniques. This paper introduces the concept of ensuring product quality right from the start, and Quality, Cost, and Delivery, or QCD, which is how we internally perform quality management.
Fujitsu mainframes have played a central role in social infrastructure systems for more than 35 years from the M series to the Global Server (GS) series, supporting customers' property succession. Its accumulated technology has been the foundation for developing other servers. And, in order to meet a wide range of changing needs, we have been developing the GS series. Recently, from the perspective of fast-paced changes in IT strategy and IT cost reduction, consolidation in large-scale systems has been accelerating. Therefore the GS series is required more strongly than ever to provide the high performance and high reliability necessary for business continuation. Important future issues are giving consideration to environmental burdens with green IT and responding to the Cloud computing era. This paper provides an overview of the hardware and software of the GS series, and its supporting technologies, and explains our approach to high performance and reliability.
By employing proven mainframe technology, Fujitsu has become a major player in the mission critical UNIX server market. Fujitsu has and continues to invest heavily in the development of features targeted at server consolidation based on virtualization technology. Its goal is to reduce customers' initial investment and Total Cost of Ownership (TCO). Along with technological improvements to further performance and functionality, Fujitsu SPARC Enterprise servers offer smaller granularity of virtualized resources leading to a greater capability for server consolidation. Overall, the SPARC Enterprise series offers the greatest opportunity for continuing to reduce TCO. Other features help reduce TCO by using a function of mixed CPUs to optimize initial investment and centralized control of servers to reduce management cost. This paper describes the various functions and characteristics of the SPARC Enterprise series, which reduce TCO throughout the ICT infrastructure lifecycle, while protecting the customer's investment and stability of the system.
Fujitsu's PC server series started to develop from the beginning of the 1990s. At the beginning its name was FM server. Since 2000, Fujitsu has been delivering PC servers globally under the brand name PRIMERGY. The hardware mainly uses Intel's architecture, and supports standard OS's like Windows and Linux, and software like VMware, all of which have a high share in the market. Thus, Fujitsu has continuously provided various products to meet a range of customer requirements. This paper explains the advantages of the PRIMERGY series, especially PRIMERGY CX1000 which is highly optimized for use in data centers. The main chapter explains the details of the architecture, its energy saving technology and cooling technology.
As we enter the age of Cloud computing, server platforms are required to incorporate virtualization as basic technology for effective use and efficient operation of ICT resources, in addition to high reliability and availability. In 2005, as a result of bringing together the know-how and experience acquired over many years of developing mainframes based on Linux/Windows and open standard IA technology, Fujitsu developed the mission-critical IA server PRIMEQUEST. This server has opened up a new market for open mission-critical systems and been used for various social infrastructure systems ever since it was brought to market. For further evolution, we developed PRIMEQUEST 1000 Series, containing new models provided with enhanced virtualization, long-term support and improved green ICT capability in addition to the features of the existing series. It went on sale in March 2010. This paper describes the aim and features of the 1000 Series and the technologies adopted.
Following on from the top selling K series of midrange computer, the PRIMERGY 6000 series of powerful business servers can be used as the main server for midsize companies or individual departments. It offers strong support to enterprise systems. In the rapidly changing business computing environment, the PRIMERGY 6000 series adapted flexibly to new technology with three significant features: high performance, high reliability, and inheritance of existing resources. Consequently, the hardware and OS of the K series, which were Fujitsu's unique architecture, could fuse with open-source technology and evolve. In this way, Fujitsu has been providing new functions constantly. This paper mainly describes the features of the OS, highly reliable technology we have developed over the years and the virtualization approach of PRIMERGY 6000 in the Cloud computing era.
IP networks have evolved to offer higher speed, higher capacity and higher safety and economic efficiency, and a variety of services are expected to be built on them in the future. Promptly making these IP network services a reality and achieving smooth network expansion while inheriting assets are important factors in the construction of future systems. The PICMG 3.0 Advanced Telecom Computing Architecture (ATCA) is an open international standard established for meeting these demands, which come mainly from telecommunications carriers. As a platform for building future IP network services, Fujitsu provides the UB300 series, which are network service infrastructure products, as well as other hardware and middleware products in compliance with the ATCA. They have already been put into use in commercial systems. This paper gives a description of the latest ATCA-compliant model, mainly focusing on a hardware platform offered by Fujitsu.
In recent years, the workload of ICT managers has been expanding steadily accompanying the increase in ICT system complexity and amount of data. Having an effective management feature is now a significant point in storage system design after having a reliable data storage feature as a basic function. And also today's society is very interested in environmental issues. Therefore how a product pays attention to the environment is a key feature of products with a functional storage feature. This paper explains the effective management feature and design for the environment that the ETERNUS DX series develops.
There are urgent needs to reduce costs aiming for system optimization in ICT infrastructure. To this end, there are high expectations for virtualization technology. By default, the Oracle Solaris Operating System offers Solaris Containers and Oracle VM Server for SPARC, which realize server virtualization, and Solaris ZFS which realizes storage virtualization. Solaris ZFS is a highly reliable file system that enables efficient disk use by storage pool technology and snapshot and clone technology. This paper covers domain management, resource management, domain migration for the Oracle VM Server for SPARC, and zone management, resource management, and zone migration for Solaris Containers. Moreover, it introduces the ZFS storage pool and ZFS file system for Solaris ZFS.
In recent years, it has become common practice to implement virtualization technology across servers, storage devices, and networks. However, simply using virtualization technology is not enough to achieve our intended purposes, which are to optimize systems and reduce costs. The first and most important thing is how to make maximum use of virtualization technology to meet the objectives. As technology advances, it is our responsibility to improve or innovatively change the respective customers' ICT environments by leveraging advanced technology. This paper discusses our strategic approach to the virtualization of the SPARC Enterprise system, Fujitsu's proprietary UNIX server, from the point of view of the technologies used. It also describes how our efforts have contributed to the ICT society, introducing some case examples.
Linux is open source software that has evolved rapidly owing to the development, tests, and bug fixes conducted by various volunteers in the Linux community. Fujitsu places Linux as a primary OS for IA server that covers a range of needs from volume-related to mission-critical-related ones. Fujitsu is actively participating in the Linux community and helping to enhance Linux. This paper describes four features being developed by Fujitsu, in conjunction with the Linux community, after the release of the 2.6.18 kernel that is the base kernel of Red Hat Enterprise Linux 5. These are 1) the Cgroup which manages system resources, 2) tracing features to improve the visibility of system activities, 3) scheduler improvement for fairness and responsiveness, and 4) utilizing the Machine Check Recovery Architecture, which has been introduced to the Xeon 7500 series recently, to minimize the effects of hardware failures and recover from failures.
To meet needs for building and operating private Clouds within enterprises, Fujitsu has globally launched ServerView Resource Orchestrator. As a resource pool that has virtualized the resources of servers, storage and networks, when operating a Cloud within an enterprise, this product automatically deploys L-Server to internal users while performing integrated management. L-Server is based on a template of the standardized logical server. Able to handle VMware, Hyper-V, Xen, which are widely used as virtual infrastructure, it is effective for integrated management of all Cloud environments. In addition, it supports the integrated operation of real servers that used the functions of ServerView Resource Coordinator VE. This paper describes Fujitsu's strategy for migrating enterprise systems to Cloud-based systems, and describes the concepts of Resource Orchestrator with an application that forms the foundation of the migration.