Dynatrace for z/OS

Dynatrace can accompany you in your IBM Z modernization journey to hybrid cloud infrastructure with end-to-end observability from the mainframe to the cloud, and everything in between. Dynatrace can also provide transactional insights into your workloads from the mobile frontend down to mainframe programs so that you can troubleshoot anomalies on the code level.

Learn how Dynatrace addresses the most typical mainframe challenges:

Is the mainframe part of the problem?

The AI-powered fault domain isolation pinpoints the root-cause of problems and assesses their user impact so that you can prioritize mitigation strategies and reduce the mean-time to repair.

All monitored LPARs, regions, and applications are contributing to this fault domain isolation.

Who is calling the mainframe and how often?

Backtrace transactions using the Service backtrace to understand your mainframe workloads and benefit from potential IBM discounts (see for example the IBM mobile and public cloud workload discounts to lower your monthly peak rolling 4-hour average MSU value).

The Service backtrace below shows how a CICS transaction interactions with both a mobile application and a web application. You can clearly see how often these applications call the CICS transaction, and also which of their requests failed.

What transactions are expensive or slow?

Analyze the performance of your transactions using the Service flow to verify if they fulfill the defined SLOs with service-level metrics. The request count for example can indicate when a transaction is called too often from an open-system, which could result in additional costs.

Use the PurePath distributed traces code-level insights to optimize your programs.

Modernize programs on z/OS for hybrid cloud?

Use Dynatrace to ensure business continuity for traditional environments by monitoring middleware technologies like enterprise service buses or message queues. While transforming your mainframe programs to make them accessible for cloud functions with z/OS Connect EE, monitored by Dynatrace.

See the end-to-end trace from z/OS Connect EE down to an IMS DL/I database below.

Set up monitoring

Dynatrace provides code modules for CICS, IMS, and z/OS Java technologies so that you can achieve seamless observability with trace and metric insights. To learn more about the supported technologies, see Technology support.

The CICS, IMS, and z/OS Java code modules interact with the Dynatrace z/OS Data Collection subsystem (zDC) via shared buffers. The zDC subsystem controls communication between the code modules and the zLocal. The zLocal, hosted in the z/OS Unix system services (USS) environment, runs as part of zDC. It handles the TCP/IP connection and transfers monitoring data from the code modules to the zRemote module.

The zRemote module processes monitoring data received from the zLocal and routes that data, compressed and encrypted, via its local ActiveGate to Dynatrace. Hence, the zRemote module offloads much of the processing work from the code modules incurred in instrumenting subsystems and applications to an open system.

To get started, see z/OS installation overview.

Licensing

Monitoring of the CICS, IMS, and z/OS Java code modules are consumed based on million service units (MSUs). Therefore, z/OS monitoring doesn't contribute to the consumption of host units or host unit hours.

For more details about licensing, see Mainframe monitoring on IBM z/OS.

FAQ

Who needs to be involved in a typical Dynatrace for z/OS installation?

To find the procedure and the people involved people, see z/OS installation overview.

How much CPU time do code modules consume while instrumenting subsystems and applications?

The code modules consume some CPU time while instrumenting subsystems and applications on IBM Z, but this overhead is typically very low (in the range of 1%-2%, depending on the type of monitored transactions). See the examples below.

Industry of customer	Country	Code module	Measured in year	Measurement method	CPU time overhead
Financial (Bank)	Spain	CICS	2020	HIS profiling¹	< 1.0 %
Insurance	Germany	CICS	2020	ran their own tests	< 1.0 %
Insurance	Germany	IMS	2020	HIS profiling¹	< 1.0 %
Financial (Bank)	Germany	CICS	2019	ran their own tests	< 1.0 %
Insurance	Germany	IMS	2017	HIS profiling¹	< 1.61 %
Insurance	Germany	IMS	2017	HIS profiling¹	< 0.33 %
Financial (Bank)	Austria	CICS	2015	HIS profiling¹	< 2.04 %
¹ Using the Hardware Instrumentation Services from IBM.

Notes

The CPU time overhead numbers are calculated relatively to the address spaces in which the code modules are running. When you compare the CPU time overhead relatively to the LPAR, then these numbers are even lower.
For example, 2% CPU time overhead in CICS address spaces represents only 1% CPU time overhead per LPAR if the CICS workloads consume only 50% of the total CPU time on a given LPAR compared to other workloads (such as jobs and system tasks).

A volatile CICS or volatile IMS is created automatically by Dynatrace when the maximum number of service IDs that can be generated per region (process) is exceeded. To increase the limit of service IDs that can be generated, please contact a Dynatrace ONE product specialist by selecting the chat button in the upper-right corner of the Dynatrace menu bar.

Can the CICS and IMS code modules capture dynamic SQL statements?

No, the CICS and IMS code modules can only capture static SQL statements.

Can I use span attributes captured by the z/OS Java code module as request attributes?

Yes, you can create Request attributes for any captured span attribute. To do so

In the Dynatrace menu, go to Settings > Server-side service monitoring > Request attributes.
Select Define a new request attribute and enter the name and data type of your request attribute.
Select Add new data source and select as Request attribute source the value Span attribute.
Enter your Attribute key.
Select Save.

To learn more details about span attributes and how to capture them, see Span settings.

Can I merge process groups created by the z/OS Java code module into a single process group?

No, process groups created by the z/OS Java code module can't be modified or merged.

As an alternative you can organize your process groups by defining metadata or defining tags based on environmental variables. Both concepts apply to z/OS Java as well.

Note at you can define environment variables only on the process level, not on the host level.

Can I use host groups to organize multiple LPARs?

Yes, you can organize multiple LPARs using host groups. For more information, see Define host groups to organize multiple LPARs.

How does the CICS/IMS code module instrumentation work?

The CICS code module uses hooks to instrument CICS terminal and application owning regions, creating events of interest.

The IMS code module uses the logging facility to instrument IMS control and message processing regions, creating events of interest from parsing binary logs.

Both code modules use hooks to instrument IBM MQ, DB2, and DL/I.

No byte code instrumentation is used by the CICS and IMS code modules.

Linux on IBM Z

With Dynatrace, you can get observability for the Linux operating systems on IBM Z and LinuxONE mainframes using OneAgent on Linux. To learn more about supported Linux distributions on the s390 architecture, see Technology support.

Monitoring of OneAgent on Linux is based on host units consumption.