AIOps on IBM Z - Group home

IBM mainframe AIOps solution and typical use case series: use case 1

  


CICS transaction performance data visualization and response time anomaly diagnosis

Background

Global market research organization IDC forecasts, by 2024, enterprises with AI will be 50% faster than those without AI in responding to customers, competitors, regulators, and partners. By adopting AI in IT operations, that is, AIOps, enterprises can automatically self-detect, decide and act on anomalies in real-time. Enterprises can avoid unexpected IT failures and business interruptions to protect corporate revenue and reputation.

Mainframe is the most important IT infrastructure for enterprises. Most critical business applications and systems are running on the mainframe, like core banking and credit card core systems. However, mainframes are still operated in a traditional way. First is the lack of visualized modern dashboards. Second is very limited analysis tools. Historical operational data is still analyzed through batch reporting. How to combine emerging technologies such as big data, AI, and cloud computing to enhance the automation and intelligence of mainframe operations has become a new trend in the transformation of mainframe operations. Data visualization and AI play key roles in speeding up the transformation to intelligent mainframe operations. A unified AIOps platform that integrates with distributed cloud computing environment is the future of AIOps solutions for mainframe. This platform is also an essential part to integrate mainframe with enterprise hybrid cloud architecture.

Through typical use cases, live demonstrations, and application workloads, these post series will show you how to leverage data visualization and AI in mainframe operations, and how to use IBM’s integrated cross-platform solutions to implement mainframe AIOps use cases and the end-to-end use cases from distributed platforms to the mainframe in the hybrid cloud architecture. These series will introduce the following typical use cases:

  • Transaction performance analysis
  • Capacity analysis and forecast
  • Application performance management and observability
  • Anomaly detection
  • Problem root cause analysis
  • Dynamic thresholds
  • Event aggregation analysis

Use Case introduction

This article will introduce the first typical use case of the series - CICS transaction performance data visualization and response time anomaly diagnosis.

Traditionally, mainframe performance data analysis was demanding on system environment and expert skills. Normally, it’s difficult for developers and operational engineers to acquire the knowledge and skills required for the whole analysis process, so it’s highly dependent on mainframe performance experts. On the other hand, the time-consuming analysis process reduced the frequency and effect of performance test and analysis. Without going through a thorough performance test and analysis, application or system changes will cause huge risks to the stable operation of the production environment.

IBM mainframe AIOps solution integrates various operational data with SME expertise from performance analysis tools (such as RMF, CICS Performance Analyzer, OMEGAMON for Db2 Performance Expert, IBM Z Performance and Capacity Analytics or IBM Z Decision Support) and uses the dashboard to visualize key performance metrics and log data in real time. The solution can compare real-time data with historical baseline, so that performance metrics and analysis results can be quickly accessed anytime and anywhere. Its concise wizard-style dashboard guides you to analyze the root cause of performance problems step by step.



This use case will demonstrate how AIOps solution can visualize CICS transaction performance data, and quickly diagnose and locate the root cause of CICS transaction response time anomalies. The architecture diagram in this use case includes five parts:

  1. IBM Z Common Data Provider: It is used to obtain mainframe operational data in real-time, such as SMF data and Syslog. Then, it transmits operational data to Elastic Stack.
  2. Elastic Stack: It is a big data analytics platform that converts, indexes, and stores operational data. It provides data query and easy-to-customize visual dashboards.
  3. IBM Cloud Pak for Data: It provides a graphical modeling environment through the built-in SPSS Modeler. You can quickly build models for historical CICS transaction performance data without programming, identify the baseline behavior of metrics, and compare the baseline with real-time performance data to identify anomalies in CICS transaction response time.
  4. Customized Elastic Kibana dashboards: Kibana provides diverse visual components and an easy way to build dashboards. Using Kibana, you can quickly build a variety of graphical dashboards. Links in the dashboard can guide you to identify anomalies from overall system health status to specific transaction performance metrics.
  5. Problem Insights Server: It is a component of IBM Z Operational Log and Data Analytics. It can automatically identify predefined anomalies and connect to corresponding system logs to help operational engineers locate problems within a short time


Demonstration

Now, let’s see a demonstration of this use case.


Use case design and development review.

Now, let’s review the whole use case design and development process.


Summary

At last, let us sum up the above use case.

Through the use case of CICS transaction performance data visualization and response time anomaly diagnosis, we demonstrated:

  1. How to use dashboards to display mainframe performance metrics in real-time.
  2. How to use AI models to build the baseline of metrics to assist operational engineers in anomaly detection.
  3. How to use data visualization technology to quickly diagnose the possible causes of anomalies.
  4. How to automatically match anomalies with predefined common errors to locate more detailed information of the problem.

This demonstration aims to show the mainframe users the business value and technical feasibility of the mainframe AIOps use cases. During the whole demonstration process, all data visualization and AI model construction can be completed without programming for performance problem analysis and root cause diagnosis. Compared with traditional methods, which rely on batch reports to analyze performance data and manually search massive system logs, it can not only greatly reduce the time and system resource requirements required, but also lower the technical threshold for performance analysis, thereby improving the efficiency and accuracy of performance problem analysis, further improve the software quality and ensure the stable operation of the mainframe production system.

These series showcase the business value and technical feasibility of the mainframe AIOps use cases. Data visualization and AI modeling in the demonstration are completed without programming for performance anomaly detection and diagnosis. Compared with traditional ways of batch reporting and manual search for massive system logs, the solution can deliver the following benefits:

  • Greatly reduce the time and system resources required for performance anomaly analysis
  • Lower the techical barrier for performance analysis and enhance the efficiency and accuracy of performance anomaly analysis
  • Improve software reliability to ensure the stable operation of the production environment

Hope the introduction of this use case can bring some inspiration to mainframe users. By referring to the solutions and technical methods used in this use case, you can develop visual dashboards and build artificial intelligence models by yourself, and apply them to more practical daily operational scenarios, to simplify the existing mainframe operation tasks, and also helps to accelerate the further integration of mainframe operations with hybrid cloud platform operations. 

We hope this use case can shed some light on mainframe operations. The solutions and technical tools in this use case enable you to autonomously build visual dashboards and AI models, which can be further applied in more operational scenarios. In this way, you can make mainframe operations simpler and accelerate the integration of mainframe operations with hybrid cloud platform operations.

Please stay tuned for more use case introduction and demonstrations of the IBM mainframe AIOps solution and use cases series. Thank you!


Watch the complete video for this use case.