Business requirements and the development of IT architectures

IT infrastructures are linked closely with business processes that are typically supported by a whole range of applications. Each application is a link in a chain, all depending on each other. For the IT infrastructure, this means that several servers with different applications run parallel to each other, they have to be available and deliver sufficient performance. If that is not the case, there is an immediate impact on the business process even causing a break in operations.

Dynamic Infrastructures addresses the needs of a modern application environment by using a pooled architecture – a pool which is managed centrally. The software is not installed on individual servers but on a central storage system from where the operating system and application software can be started on any server in the pool – on physical and virtual servers alike.

Such central management of the pool also covers high-availability functions: If one server fails, the affected services are automatically started on another server in the pool.

Dynamic server infrastructures offer multiple benefits:

• Pooling servers greatly reduces excess capacities. System environments can be consolidated and utilization optimized.
• Entire business processes are protected by a single high-availability concept. At the same time, less critical applications can be operated reliably at very little extra cost.
• The dynamic assignment of servers to applications enables rapid roll-out of new applications.
• It also simplifies maintenance, reduces downtimes and permits rapid response to unplanned workloads.
• A pool can be built with a small number of servers and expanded gradually. Unlike large system architectures, the up-front expenditure on initially unused capacities is lower.

An ideal dynamic infrastructure is built on the following core components: 

1. Blade servers
Blade servers are an ideal format as they are themselves a pool architecture. Their design alone already produces a consolidation effect, for example regarding I/O connections, power supply and cooling. Centralized operation results in reducing administration overheads.

2. I/O virtualization
One requirement for the flexible assignment of applications to servers is a virtualization layer between the servers and the storage area network (SAN) as well as the local area network (LAN). This is due to the fact that, without I/O virtualization, any change in a server's allocation would require many manual steps and very close coordination between the server, network and storage administrations.

3. Server virtualization
The power of the latest server generation is sufficient for just about all workloads. In order to improve the degree of utilization, servers are being virtualized more and more by installing hypervisor products, such has VMware ESX, Hyper-V and Xen. A large number of virtual machines can thus run on one server and some unique server virtualization features (e.g. Live Migration, etc.) can be used.

4. Centralized storage
A prerequisite for dynamic infrastructures is a centralized storage system which is needed for the dynamic assignment of servers to applications. The storage system hosts all software images – the operating systems, the applications and the files for the virtual machine. In order to guarantee reliable and high-performance operations, storage must be highly scalable in terms of capacity and throughput, but should also provide functionality for data availability and integrity.

5. Dynamic resource management
A wide range of resources have to be administered in a server pool – physical servers, virtual servers and their SAN and LAN addresses. In addition, storage capacity has to be allocated to every server. Central management via a resource manager is necessary to enable the dynamic administration of all these resources and thus the efficient operation of a server pool. It can be used to reduce the time and effort involved in having to coordinate every task with server, network and storage specialists. The organization concept here is that storage and network administration supplies quotas with storage capacities and network addresses to the server administration which manages them with the resource manager. This technology makes it very easy to implement high availability for the entire server pool on an automatic basis. A large number of production servers are protected by a handful of substitute servers.

Fujitsu has developed hardware and software products that seamlessly integrate to shape a dynamic infrastructure. All the components are pre-tested in combination to ensure a fast and risk-free implementation.

PRIMERGY BX900
The PRIMERGY BX900 blade server system has been specifically developed to enable efficient operation of a server pool. A lot of computing power in a very small space, maximum main memory configuration and high I/O bandwidth means that all types of workloads – on both physical and virtual servers – can be centralized. Consolidation of network connections, integrated power management and central control of all physical components comprehensively optimize the pool architecture.

ServerView VIOM
Separates the network identity of a server from the hardware. It virtualizes the specific I/O parameters of a server and makes them available in hardware-independent server profiles

ServerView Resource Coordinator VE
Provides uniform control functions that can be used for physical and virtual servers. New servers are added if workloads increase or new applications are launched. Servers are replaced and applications relocated from one server to another as resource requirements or workloads change. If a server fails, ServerView Resource Coordinator VE automatically starts a spare system to prevent downtime. A large number of production servers in a pool can thus be protected by one or a few substitute servers (N:1 failover).

ETERNUS DX400 / 8000
ETERNUS DX400 series and ETERNUS DX8000 are disk storage systems that serve as a foundation for dynamic infrastructures. They deliver the functions for scalability, high-availability and data integrity that are needed for the reliable operation of pooled environments where all system images are stored centrally. Examples are a modular architecture, high redundancy of key components, validity checks to prevent data corruption, automatic rebuild after drive failures, and the latest raid technology with redundant copy and so on.