PRIMERGY BladeFrame: Architecture and functional principle

What enterprise wouldn't want to have IT that can hold its own with the business at all times? New business opportunities could be leveraged more quickly, the day-to-day business would be sure to always have optimal capacity, the effectiveness and efficiency of business processes would improve, and more economical IT would make an important contribution to the success of the enterprise.

With PRIMERGY BladeFrame powered by Egenera, Fujitsu Siemens Computers clears the way for highly responsive enterprise IT. PRIMERGY BladeFrame BF200 and BF400 are identical in construction to the respective Egenera servers Bladeframe ES and EX and therefore are 100% compatible to each other.

The solution features a service-oriented infrastructure, the innovative Processing Area Network (PAN), which radically eliminates the complexity of traditional infrastructures. Thanks to the simplification, separation and sharing of resources, IT departments have experienced enormous gains in agility and efficiency. Applications can be deployed within a matter of minutes, instead of weeks, and investment costs and the total cost of ownership can be demonstrably reduced by 50 percent.

PRIMERGY BladeFrame is a service-oriented infrastructure in which processing, network and resource management are integrated and ready for use. The best of two worlds is combined in PRIMERGY BladeFrame - the kind of administration functions that are valued in the mainframe world with a blade-type processing architecture, based on industry standards. This makes the solution a "Dynamic Data CenterTM in the box" that goes far beyond traditional blade architectures and presents a completely new approach to utilizing IT.

The integration of the PRIMERGY BladeFrame PAN architecture in the Dynamic Data CenterTM strategy is another stage on the road to a new generation of IT infrastructures. The significant improvements they bring about for IT operations in key areas are evidenced by the Dynamic IT solutions from Fujitsu Siemens Computers being successfully deployed. Furthermore, they make clear that IT departments are completely redefining their position in the enterprise with the gradual implementation of the Dynamic Data CenterTM. They actively shape business processes and can utilize considerably more resources for innovation, thanks to extremely economical IT operations.

PRIMERGY BladeFrame: Architecture and functional principle 

PRIMERGY BladeFrame organizes the resources into a Processor Area Network (PAN). The pBlades serve as processor units, sBlades act as an internal switching center for the communication among the pBlades via an internal high-performance network, and cBlades serve as the control center and host the actual brain of the solution, the PAN Manager software. The PAN architecture thus makes PRIMERGY BladeFrame a service-oriented infrastructure for the next-generation data center:

• Radically simplified infrastructure overcomes complexity and the related costs and obstacles in IT operations
• Resource sharing through a wide variety of applications
• Flexibly scalable to meet performance needs and business requirements for the availability of IT-based business processes

pBlades (processing blades)
As already outlined, the PRIMERGY BladeFrame computer nodes have been scaled down to the bare essentials and are stateless servers with CPU and memory resources. Every pBlade can be equipped with up to four industry-standard processors. A BladeFrame chassis has room for up to 24 pBlades and 96 processors. In other words, up to 24 virtualized servers for applications running on Windows, Linux and Solaris are completely at the customer's disposal in a chassis. pBlades with varying resources (main memory, a number of processors, processor clock speed) and pBlades with Intel processors, as well as pBlades with AMD processors, can be operated in a chassis without any problems. Every pBlade is connected to two sBlades via a redundant, internal high-performance network.

sBlades (switch blades)
Two sBlades, together with the serial bus, form the internal network in the PRIMERGY BladeFrame. One sBlade controls the pBlade-to-pBlade communication as well as communication to external instances (LAN/SAN). If one sBlade should fail, the second one takes over. The connection set-up between the pBlades and the physical switch unit of the point-to-point PAN architecture is dynamically created.

cBlades (control blades)
Two redundant cBlades make up the control center. They host the PAN Manager, PRIMERGY BladeFrame's integrated virtualization and management software, and they are also the real I/O connection to LANs and SANs. Each of the cBlades has multiple 2 Gbit/s Fibre Channel host adapters and Gigabit Ethernet adapters. During normal use, the two cBlades handle the incoming and outbound traffic, and optimal use is made of the bandwidths. If one cBlade should fail, the second one takes over its function.

PAN Manager software
This is the brain of the entire processor area network. PAN Manager is responsible for virtualizing servers, allocating resources and completely managing the PRIMERGY BladeFrame solution. A virtual pServer is created by defining a pBlade with assigned disk storage devices and network resources. These virtual pServers are also managed via PAN Manager. Administrators can add, switch, or change pServer definitions in the entire server pool and do the same with the virtual I/O interfaces. Furthermore, PAN Manager monitors the respective allocation of LUN storage areas to individual pServers. Once a pServer has been defined by PAN Manager, an image of the operating system and the application are deployed on the pServer and monitored.
What's more, PAN Manager has a function layer for the implementation of failover high availability at the hardware and application level. Failures of a pBlade are automatically identified, and the identity of the pServer concerned, along with the associated (virtual) resources, are mapped to a spare pBlade and the operating system and application are automatically restarted. This entire process is carried out at the software level so that an application is available again within a matter of minutes. If there is not a spare pBlade, PAN Manager offers functions for starting/stopping and shutting down production applications according to predefined rules. Depending on predefined thresholds for network load, CPU utilization, memory utilization, etc., PAN Manager can also automatically create new pServers without having to reboot the system.

If you want to set up a secure, and possibly also a high-availability data centre nowadays, then you will need a strong partner....

Latest News