For far too long, the mainframe has been treated as an isolated citizen from a broader Application Performance Management (APM) perspective. As a result, IT has had little visibility into the impact of distributed applications’ transactions on mainframe transactions and workload, hindering the optimization of mainframe application performance. This leads to poor user experience, inefficient use of mainframe resources and a lot of wasted time and energy spent chasing down the source of performance problems.
But with the increasing demand from Web and mobile applications, requirements for superior mainframe application performance have surged. This is because the mainframe increasingly supports business-critical, customer-facing applications in industries such as banking, retail and telecommunications, to name a few. IT teams must have a way to quickly identify and resolve poorly performing mainframe transactions in order to increase user satisfaction, reduce costs and support business growth.
The key is having a complete end-to-end transaction monitoring and APM solution that spans the entire application infrastructure—from user click, across all application tiers and through the mainframe application. Continuous, real-time visibility into all mainframe CICS, Java and WebSphere Application Server (WAS) transactions can be essential to helping businesses better manage mainframe application performance. End-to-end transaction management from the edge of the Internet to the depth of the mainframe can help distributed system and mainframe teams resolve performance problems faster, reduce MIPS costs, postpone hardware upgrades and accelerate time to market.
End-to-End Transaction Management Available for z/OS Environments
End-to-end transaction management solutions are available in several variants, including for CICS and Java applications. These solutions can identify hotspots across distributed systems, including where I/O, synchronization or high CPU time impact system performance. Transactions are traced through messaging connector technologies such as WebSphere MQ, Message Broker (MB) and Enterprise Service Bus (ESB) that initiate the CICS or WAS application transactions. User experience management capabilities can also be leveraged to visualize if users are happy, tolerating or frustrated with their experiences; and demonstrate the business impact of slow performance on site usage and revenue flow.
All transactions that flow across CICS regions can be traced, including DB2 interactions. Individual session information is captured for all applications operating within a CICS region or within multiple regions within a CICSplex, including inter-process communication messages across CICSplex regions. Comparable visibility is provided into WAS transactions.
Integration With Application Performance Monitoring Systems
Application performance monitoring and analysis solutions for mainframe applications can pinpoint application inefficiencies such as an application making too many database calls, which may cause excessive CPU and MIPS consumption. IT departments can then make modifications to application code to reduce hardware and software costs while increasing user satisfaction. The activity of both online and batch processing applications in single and multiple system environments can be measured.
End-to-end transaction management solutions can then be tightly integrated with application performance monitoring solutions to provide a complete means for managing mainframe transaction and application performance. For example, when a performance issue is found, it’s possible to navigate directly to a dashboard to view session measurement data that reveals causes of excessive mainframe CPU consumption and wait time. Users can easily drill down into COBOL, C/C++ and PL/I source code for root cause analysis as well as determine ways to increase the performance of DB2 SQL statements, reduce wait states and eliminate resource overuse. Application performance monitoring solutions can also automatically display lines of existing source code and provide recommendations for actually improving application efficiency and responsiveness.
This total-picture view provided makes it much easier for organizations to locate and correct application performance problems and inefficiencies related to code and logic. Subsequently, this helps organizations make more informed decisions when managing CPU consumption and hardware costs, as well as anticipate and resolve application performance issues. By integrating end-to-end transaction management solutions with application performance monitoring solutions—and proactively managing both mainframe application and distributed system performance—organizations can achieve better application performance combined with greater balance throughout the entire end-to-end transaction path.
Accurate, Continuous, Real-Time Information to Address the Full Application Lifecycle
End-to-end transaction management traces and captures 100 percent of all transactions, delivering precise, accurate, real-time transaction information—not sampled performance information. Other approaches use sampling and analytics to approximate mainframe application performance, rather than continuous monitoring. This means that if a performance issue occurs outside the sampling interval, the performance problem won’t be found until much later and after one or more users experience frustration with their application performance. Sampling leaves visibility gaps in the application performance continuum, no matter how intelligent the smart agent is.
Transaction paths are automatically generated the moment smart agents are injected into mainframe applications and they monitor all the application transactions that take place. There’s no further configuration or mapping needed or code hooks or methods to be applied. End-to-end transaction management delivers needed visibility and key insights throughout the entire application lifecycle. Potential performance problems can be identified and fixed during the development process, and scaling and other performance issues can be identified under load, in test/QA and development before applications are sent to production for deployment. Once an application is in production, end-to-end transaction management continues to offer critical insights that support faster Mean-Time-To-Repair (MTTR) and user satisfaction.
The Power of Analytics
Transaction details can provide a series of built-in powerful views and analytics for mainframe and distributed system components. These features take the large quantity of data generated and can make it usable, actionable and insightful for mainframe and interconnected application architects, developers, testers, performance specialists and operations teams. Every transaction can be reviewed against preset thresholds and patterns to alert operators of hotspots and breaches, leading to more proactive problem identification and triage.
For a quick determination of what has changed from version to version or what’s different between transaction iterations, end-to-end transaction management has a built-in compare capability that quickly analyzes the associated transaction paths and determines what attributes are better, worse and new. It can then display those results in a readily recognizable format in a dashboard.
For application testers, analytics are applied to connect transaction paths to load test application trouble spots. This makes it possible for test teams to pass sessions to the application developers, complete with integrated logs combined with load test reports. For every slow performing page and component in the report, testers can provide, and developers can drill into, transaction path details in order to fix problems and significantly reduce or eliminate test reruns.
Analytics can automatically detect application performance hotspots for application performance teams, such as the longest running transactions, high number of database calls per request, memory leaks, duplicate threads, chattiness among tiers and services and others. These hotspots show the full root-cause information as well as all the applications that pass through the hotspot.
Today, IT departments can proactively manage distributed and mainframe systems as a single entity. The result is a comprehensive, continuous transaction-oriented APM solution for the entire mainframe application infrastructure, assisting development, test/QA, as well as production/operational teams in their quest to find and resolve application performance problems. Improved visibility into what’s driving mainframe workloads results in significantly reduced MTTR, improved mainframe cost management and increased customer satisfaction. Mainframe resources are better optimized. Application users are happier. IT operations run smoother. And the business IT infrastructure operates at a higher level of efficiency.