Not logged in » Login
Jul 05 2021

Anomaly Detection for SAP Landscapes

A newly released anomaly detection approach couples machine-learning techniques with domain expertise in order to create best-in-class customer experience. Deep dive into your SAP operations using machine learning ensembles with Fujitsu’s SystemInspection Service for SAP Solutions.

Image

 

Anomaly detection for SAP landscapes is now available as part of the SystemInspection Service for SAP Solutions standard offering. The mechanism makes it possible to identify periods of irregular SAP operations with a special focus on performance. As the underlying root causes can potentially impact business continuity, it is extremely important to detect and investigate anomalies as early and thoroughly as possible. The developed mechanism leverages a machine learning ensemble approach that results from joint research activities of Fujitsu and the Otto von Guericke University Magdeburg, thus combining Fujitsu’s long-standing domain expertise with state-of-the-art research artifacts.

 

The Fujitsu SystemInspection Service for SAP Solutions

The SystemInspection Service for SAP Solutions is a packaged offering by Fujitsu, which aims at measuring and analyzing SAP environments by combining data mining techniques with comprehensive consulting expertise. It addresses use cases related to SAP infrastructure related in the field of transformations, assessments, or trend analyses in a neutral and data-driven manner. The goal is to get a complete understanding of the current workloads, performance, and related resource consumption. In this way, measures and guidance will be provided to optimize your SAP landscapes in accordance with the business strategy and requirements.

The Fujitsu SystemInspection Service for SAP Solutions creates a holistic inventory of complete SAP landscapes. It collects and analyzes numerous performance metrics covering physical and logical layers such as infrastructure components, transactions, or dialog steps. The measured information serves as a comprehensive foundation to provide data-driven decision support for any upcoming transformation, migration, transition, or hardware refresh project. In short, with the help of the SystemInspection service, one can:

  • Avoid over-provisioning and unnecessary investments in IT equipment;
  • Increase the quality of services by helping to eliminate or circumvent performance bottlenecks;
  • Deliver individual and short-term results with minimum effort for the customer;
  • Create an insightful basis for capacity planning of upcoming new requirements to make strategic decisions;
  • Give clear recommendations for optimizing the IT infrastructure for SAP landscapes.

The service works with the hardware of any vendor, whether on-premises or off-premises environments. Furthermore, it is fully GDPR-compliant as only technical system statistics are processed and neither personal data nor business data are collected.

 

Decision Support

The SystemInspection Service for SAP Solutions provides clear recommendations and supports decisions in a data-driven manner. Thus, the service definitely does not end with the outcome of some classification based on machine learning. In fact, results must be interpreted and further investigated by domain experts. To support this phase, our team of performance experts makes use of the SystemInspection data discovery portal, which provides various means to analyze the measured data and deep dive into points or components of interest.

The phases and views of any consultative investigation highly depend on the use case addressed by the service. A typical use case, which benefits from our newly released anomaly detection technique, is the root cause analysis of performance bottlenecks.

If performance problems are reported by end users, any additional information can be beneficial in finding the root cause – e.g. details on the transaction associated with the issues, the exact time and date, and information on preceding activities. However, in many cases, problems are described rather generically and user complaints do not point to a specific event. Therefore, the initial challenge is typically to identify any system activity, which has shown performance degradation compared to periods of normal operation. Secondly, these must be linked to exact timestamps in order to allow in-depth performance and workload analyses, reaching down to transactions, background jobs, or processes that co-existed at that time. The SystemInspection anomaly detection mechanism supports these steps by scanning all measured data points and separating the normal from the anomalous. Results are visualized in an interactive 3D chart, which is part of the SystemInspection data discovery portal. In this way, performance analysts are given the tools necessary to zoom in on and explore the data points efficiently.

 Image 

In the example case (see Figure above), a productive SAP system, formed by eight application instances was measured for a period of three weeks. The anomaly detection mechanism revealed eight points of irregular performance at 3:00 pm, one for each application instance. Mean response times jump up to more than ten seconds as compared to normal response times below one second. This clearly indicates a system-wide issue. With the output information on the exact timestamp, further investigations are well targeted and the amount of data to analyze is reduced from more than 500 hours of operation to just one hour of interest.

In the demonstrated example, our customer was able to react to the performance issues quickly and consulted the SystemInspection team to identify the potential root cause. In other cases, existing minor bottlenecks may be unknown until effects become visible in a chain of faulty events. Therefore, it is extremely important to pay attention to anomalies as these may indicate bottlenecks or misconfigurations, which could lead to severe problems or – in the worst case – have the potential to interrupt business.

Please find further details on Anomaly detection for SAP Landscapes in our respective white paper: https://sp.ts.fujitsu.com/dmsp/Publications/public/wp-systeminspection-sap-anomaly-detection-en.pdf.
 
The SystemInspection Service for SAP Solutions addresses SAP infrastructure-related topics in a data-driven manner. To learn more, please contact expert.sap@ts.fujitsu.com.
 
 

Dr. Hendrik Müller

 

About the Author:

Dr. Hendrik Müller
close share field

Lead Solution Architect for SAP with a focus on innovations around hybrid cloud and SAP operations.

Besides 15+ scientific publications in the field ...

SHARE

Comments on this article

No comments yet.

Please Login to leave a comment.

X

Please login

Please log in with your Fujitsu Partner Account.

Login


» Forgot password

Register now

If you do not have a Fujitsu Partner Account, please register for a new account.

» Register now