With the growing adoption of virtualization technologies, vendors are not only responsible for the design and sizing of physical machines for a SAP landscape but also for virtual environments.
The SAP landscape is designed to run the production model of SAP core modules. It comprises the client-server interface wherein the users interact with the core application using either a Web portal or direct communication with the virtual application server. The dialogue instance, central instance and core database are hosted on this particular server for the user. This entire SAP landscape design architecture depends on parameters such as memory utilization, resource allocation and decision making.
1. Avoiding assumptions
Business decisions for IT are not always taken by technical experts in organizations today. Such decision makers depend heavily on partners to provide them with information and suggestions on various solutions. In case there has been no detailed study of the requirements for the SAP landscape design, the decisions will be based on assumptions. This is not the best approach. Rather, some level of analysis needs to be done before designing the SAP landscape.
You can carry out the “systems requirement study” usually done by applications vendors or make sure that your SAP partner does it. This study will help you understand the current business requirements and workloads. These can then be easily mapped to the respective SAP modules during the SAP landscape design.
Another tool that can be leveraged is the “solution manager”. This provides industry-wise best practices for SAP landscape design. These are available in the SAP service portal known as “marketplace”. Once you download the best practices for your specific business vertical, you will be presented with handy information on how to manage your workloads, check the processes involved, and understand your application behavior when it integrates with all the subsystems.
2. Ensure proper memory and resource utilization on VMs
All virtualized servers require additional memory during peak load, and thus the utilization of the server is higher due to high transaction levels. You must tweak the memory for proper utilization of memory when you design the SAP landscape for a virtualized environment, in order to prevent performance bottlenecks. One way to accomplish this is by allocating static memory configuration to each virtual server. You can create a policy to dynamically reload memory to a particular server as and when it is required.
Traditionally, virtualization has been designed to give better capacity utilization. SAP on the other hand focuses on performance rather than capacity. Even if the server is running at 30% but performance is satisfactory, that is all that counts for the SAP landscape. It’s no use running the servers at 60% if the desired performance level is not achieved. Today, however, there are several tools available to determine performance requirements for each module according to the number of users accessing the module.
Quick Sizer is one such SAP online tool that will help you design your SAP landscape around your user base. Analyze the transaction load from the first day you implement the SAP landscape. You could divide your users among low-end, moderate and high-end categories according to the number of queries they run in a day. These queries have a direct impact on system load. You also need to classify the different users according to the modules they access. A module-wise table is provided by Quick Sizer, to be filled by the administrator responsible for SAP landscape design. Both these parameters will help you assess your requirements optimally.
Once you define your users and pick the modules specific to your business, you will need to feed the Quick Sizer tool additional information on your OS, database configuration, and so on. The Quick Sizer tool will then provide a report on transaction loads, CPU utilization, memory and IOPS required. You can use this information to design your SAP landscape effectively.
3. Avoid shared resources for SAP modules
Shared resources for SAP modules on a virtualized platform may trigger performance issues. Ensure that your SAP landscape does not follow a common protocol with the other systems in your infrastructure. Instead, allocate dedicated servers and storage for SAP, lest the bandwidth for SAP be reduced.
When running a mission-critical SAP module in a virtualized environment, you need to separate certain memory resources for peak loads. You can achieve more stable performance if you use memory reservation technologies.
About the author: Avik Rudra handles APJ Enterprise Product Management, SMB at Dell. Rudra has been in the IT industry for more than 14 years. His experience spans IT technology and solutions consulting for integrated database applications and building landscape for large data centers deploying enterprise solutions across India and abroad. His key focus has been on capacity utilization for IT consolidation and virtualization, with a domain expertise in SAP deployment across the country.
(As told to Mitchelle R Jansen)
This was first published in December 2011