Many systems in a high-availability SAP environment require only a fraction of the performance that current server systems offer. There are two possibilities for optimising the use of resources: Either implement several systems on one operating system instance or place several operating system instances on one server.
To minimise side effects, the second option is common practice. Suitable technologies are partitioning and virtualization.
Since all of the business processes implemented on SAP depend on the availability of the related SAP system after an SAP introduction, availability – besides performance – is of vital importance. If the SAP system is “down,” the business processes mapped there can no longer be carried out. Suitable technology approaches are called for in order to implement the requirements with regard to recovery time objective (RTO) and recovery point objective (RPO).
One approach for increasing availability is the use of cluster solutions. Alternatively, shadow databases are often used to safeguard against logical errors. Such solutions usually have specific requirements for the configuration of the operating system, since the application concerned “moves” from one operating system to a second using a suitable procedure. Operating system virtualization and hardware partitioning are also alternative solutions.
Should there be a hardware failure, the operating system can “move” the application to new hardware. Since an OS is “moved” here, an adjustment of the configurations between the two operating systems is not necessary, which is a valuable advantage from the administrative perspective.
Objectives and Considerations
The further planning for virtualization is therefore driven by technical as well as economic factors. The specifications for simple, easy-to-administer high-availability solutions are the primary concern for many IT managers. Hardware failover and virtualization offer the advantage that the high-availability solution does not have to be integrated into the operating system.
The application, operating system, network and storage levels are well separated. The use of virtual IP addresses, which must move from one operating system instance to another, is not necessary. In addition, in some cases restoration within four hours or less must be guaranteed at times. The failure of a data processing centre should not lead to the unplanned downtime of the productive SAP operation.
The Compute Blade infrastructures certified for SAP from Hitachi based on the x86 platform with LPAR virtualization therefore enable an N+M failover – otherwise only available for mainframes and high-end Unix solutions.
The Compute Blade technology developed for mainframe technology has been in operation in Japan for about six years and is the market leader in the company’s homeland. As the largest SAP distribution partner or integrator, Hitachi is also the leader for SAP applications in the land of the rising sun.
For hardware-dependent high-availability scenarios without cluster software, Hitachi Compute Blades offer integrated N+M failover. This makes it possible to define a random number of N blade servers as “cold standby” for an equal random number of productive blades; when necessary, these can be replaced by a standby blade independent of the chassis, rack or location. Besides the cluster software, users also save on hardware, which otherwise would have been necessary for a 1+1 or N+1 cluster.
Hitachi Virtualization Manager (HVM)
Beyond this, the LPAR (logical partitions) technology adopted from the mainframes with the Hitachi Virtualization Manager (HVM) also offers hardware-dependent server virtualization, which makes possible direct access to the server periphery and host bus adapter. Naturally, the original file systems from Windows and Linux are supported.
This saves users not only licence costs for the virtualization software but also it circumvents the virtualization layer with its proprietary file system. Beyond this, there are performance and security advantages for the operation of high-performance and critical applications as well as database applications. At the same time, all Compute Blades from Hitachi are certified for VMware and Hyper-V so that their use under all leading hypervisors as well as under HVM is possible at any time.
LPAR technology and dynamic-dedicated resource sharing
Another advantage of the Hitachi Compute Blades is the possibility of common use of resources, which are made available on the basis of Hitachi LPAR technology. Through this, CPUs and memory can be assigned both as dedicated for as well as dynamically to the LPARs, virtual instances or applications. Mixed operation is also possible. This not only increases the performance and security of applications and databases but also allows a much-improved use of the server hardware, which in turn further reduces investment and operating costs.
Together with the energy-saving function of the Hitachi servers and the optimum performance and I/O rates tuned to the Hitachi Virtual Storage Platform (VSP), significant and lasting cost and benefits advantages result for the failsafe SAP operation.
Results and Savings
Through the use of Hitachi Compute Blades and the application of Hitachi LPAR and HVM, both small and very large SAP systems can be populated efficiently. Operation is easy and effective thanks to the clear separation of each layer. This includes the following points:
▪ Handling clarity
▪ Excellent scalability
▪ High security
▪ Efficient use of system resources
▪ Small footprint
▪ Energy savings of 25 per cent or more
▪ Licence cost savings
In this way, availability can be optimised over several locations – for a simple and efficient management of SAP environments.