Imagine a computing infrastructure that could automatically adapt inline with the ebb and flow of day-to-day business service requests. The value of a fluid infrastructure that dynamically adjusts to both predictable and unforeseen business events could be immense, particularly if it could react to these situations in real-time and additionally reduce overall TCO.
Where hardware and software performance issues slow down business agility there are often only two choices: buy more hardware or adapt the available resources by implementing an adaptive computing approach. In the context of SAP technology, business solution service requests – at least within an Enterprise Services Architecture – should run at any time, in any place and on any server. This paradigm connects well with optimizing idle hardware and enabling customers to dynamically assign load to available resources, when and wherever required. As such SAP provides a technology called the Adaptive Computing Controller (AC-C) to co-ordinate this kind of activity. AC-C is part of SAP NetWeaver.
Maintaining optimum systems performance
The AC-C or “abstraction layer” can decouple and reinstate the relationship between applications and hardware, enabling it to start, stop and resend or relocate load to any predefined resource. During peak load times, for example the quarter close, this can help to maintain optimum systems performance as resources that would otherwise be idle are used. By adapting the available infrastructure to business needs it is possible to intelligently manage both application performance and overall IT costs.
The true power of such an adaptive infrastructure is that the applications, in conjunction with a management layer, running on the platform, determine themselves, which processes need to run, where and when. Therefore, the platform itself does not determine the load. Key to this is the management layer that interfaces between the application and the platform. SAP offers its own management layer in SAP Solution Manager and the Computing Center Management System (CCMS) that provides the system performance metric data necessary to make decisions regarding workload starting, stopping and relocating. In essence it is the integration of the infrastructure with the applications it is running that is critical to enabling a truly adaptive computing infrastructure.
Hardware must be “adaptive ready”
In order to achieve an adaptive computing framework for managing the SAP environment, the available hardware resources must first be “adaptive ready”. Furthermore, to automate adaptive computing the AC-C must connect not just to hardware but to a certified automation engine via the AC-C certifiable interface. One such technology already available is the Unisys and Redwood Software SAP compliant adaptive computing solution. The co-developed solution is certified by SAP and focuses on the ES7000/ONE server concept from Unisys, and the Cronacle for SAP solutions technology from Redwood Software. The basis functionality is the Sentinel server management software which is available with every ES7000 and allows additional servers and CPU resources to be dynamically added to requesting applications, generating an adaptive ready environment. The Sentinel application executes actions triggered by Cronacle for SAP solutions based on real-time alerts and metrics received from the wider SAP and non-SAP environments. Additional intelligence is supplied since the Sentinel application and AC-C are together enabled to automatically react to process chain dependencies or high-load limits elsewhere in the IT landscape. ES7000 hardware is therefore able to adapt, by managing its resources pools (both adaptive and conventional) on-demand and in real-time.
The unique Unisys approach is that the ES7000 does also deliver server consolidation. The ES7000 consists of a large number of separate CPUs that can be grouped together into nodes. The advantage is that these nodes can be dynamically modified to utilize less or more CPUs as workload demands. This delivers the flexibility required to reduce the numbers of machines managed by the operations team, further reducing TCO. When using Linux as the operating system, these adjustments between load and resource can also be made on-the-fly. On-the-fly, real-time adaptation to business change is ultimately the defining premise behind adaptive computing.
Aligning resources proactively
Another benefit from uniting Unisys and Redwood technologies is the added intelligence that SAP Solution Manager and the AC-C can leverage to perform complex monitoring and automation related tasks. Meaning that, stand-alone SAP Solution Manager can highlight the results of process monitoring (like status and statistics) yet it still requires someone to investigate and manually determine the next course of action. In addition, the reaction needs to be manually determined but also manually initiated, slowing down the process of adaptation and opening the door for user input error(s).
Correspondingly, exception handling parameters in Cronacle for SAP solutions allow it to initiate a response based on pre-defined escalation procedures, triggering necessary reactive processes on behalf of SAP Solution Manager (via the Adaptive Computing Controller) within the Sentinel application. Furthermore, Cronacle for SAP solutions can react to current platform and application metrics as well as expected future load. As such and due to its integration with the AC-C, load can be freed up and reassigned prior to those tasks being started. This allows for pro-active and predictive adaptive computing that aligns resources in advance of expected load.
By teaming together Unisys and Redwood Software have delivered an out-of-the-box automated adaptive computing solution that is certified by SAP and has deep integration at all levels of the SAP NetWeaver application stack. This enables the monitoring of application processes to be acted upon at the hardware level, automatically, in real-time and with predictive capabilities. This bridge between hardware and applications ensures a smooth transition between stopping, starting and relocating workload. Such flexibility provides a mechanism for guaranteeing service levels (as application workflow output becomes faster and more efficient), optimizing infrastructure ROI and further reducing TCO by minimizing administration staffing costs and related errors. Ultimately as hardware purchases can be based upon average load rather than peak load, SAP customers can save significantly by consolidating hardware and doing more with the existing resources at their disposal.