Performance Evaluation after Upgrades and Conversions with SAP EarlyWatch Alert

[dirk_jenrich]

When you are responsible for operating your SAP systems, you are surely familiar with SAP EarlyWatch Alert. It’s an automatic diagnostic service for the essential administrative areas of an SAP system, creating alerts on critical situations and offering guidance to improve performance and stability. SAP recommends that you consume the corresponding reports in the cloud-based SAP EarlyWatch Alert apps in SAP for Me.

As of now, the SAP EarlyWatch Alert Workspace (your central landing page for your system landscape regarding stability, configuration, hardware utilization and performance) contains a new card called Event Statistics.

This probably leaves you with two questions: What are events and why are they interesting to me?

Well, for the first question: events are significant changes in the technical boundary conditions of a system, such as SAP S/4HANA Conversion (which is a migration by upgrading from an existing ERP system to SAP S/4HANA), SAP Product Upgrade, SAP HANA Migration or SAP HANA Upgrade. The first two types of events are already implemented in the first version, SAP HANA Migration (see SAP OS/DB Migration Check) is active as of 15.09.22, and SAP HANA Upgrade is in the pipeline and will be delivered in one of the next versions. You monitor for sure always the corresponding systems after such a change particularly intensively and are interested in the following aspects:

• How did the system performance and system load change?

• Are there any regressions?

• Are your customizations and modifications still working as expected?

The answers to all these questions are now at your disposal in the Event Statistics. After an upgrade or a conversion, the change in the general system behavior regarding performance and load as well as particularly relevant changes during certain business hours are displayed here. This is done by comparing the median of the measured values up to eight weeks before the event with the weekly averages of the first eight weeks after the event.

After this introduction, let’s have a closer look at the content of the Event Statistics:

After examining the card, let’s dig deeper. When you click any of the elements in the card mentioned above, an event list with further information is displayed (you can also reach this list directly using the link Dashboard: Events of the last two weeks).

In this list, you can navigate to the corresponding views in the SAP EarlyWatch Alert Dashboard of the respective system for further analysis:

For the systems in scope, besides administrative information and details about the corresponding upgrade or conversion, you can see how the event changed the most important performance and load metrics. As an example, let’s have a look at some of the displayed metrics:

KPI	Description
Dialog Steps	Here, the change in dialog steps of the selected task type before and after the event is displayed (in this case, the task type is RFC - by default, the task type with the largest total load is selected): In the first line, the number of dialog steps, averaged up to 8 weeks after the event is displayed. In the second line this value is compared with the average value of the sessions for up to 8 weeks (depending on the availability) before the event. Click the number of dialog steps for a system to navigate to the load detail view in the SAP EarlyWatch Alert Dashboard of that system for further analysis.
Average Response Time	Here, the system performance change in the first 8 weeks after the event is displayed. The comparison is based on average weekly performance values. The median of these weekly values before the event is used as the basis of the comparisons (minimum 3, maximum 8 weeks). This value is compared to the weekly averages for the first 8 weeks after the event and the change in percent is displayed in 8 bars. In the tooltip, the numerical values of the changes are displayed. The above example makes the sometimes astonishing improvements of an upgrade for the system performance transparent. After a click into the SAP EarlyWatch Alert Dashboard you can see the change of the dialog response time due to the upgrade:
Severe Slowdowns	A significant performance degradation during specific hours (e.g. 09:00 - 10:00 every Monday) often indicates that reports or similar functions that are always executed at same times have been potentially impacted by an event and further investigations are needed. Such an occurrence is called a Severe Slowdown. If you are interested, the exact requirements for a severe slowdown during a specific business hour are: Daily dialog steps are at least -50% lower than the history median. The hourly dialog steps are at least -50% lower than the history median. The average response time is at least 200% higher than the history median. The average response time exceeds 1.8 seconds. The number of dialog steps per hour in the history median exceeds 5000. Note that severe slowdowns are only reported for task types that have an direct impact for the response time experienced by the end-user - in detail DIALOG, HTTP, HTTPS, and WS-HTTP. Here, the number of severe slowdowns after an event is displayed. Click the number to display date and time of the single occurrences. Click the time slot to navigate to the system load detail view in the SAP EarlyWatch Alert Dashboard of the corresponding system and task type during that week. In this case it looks like that every Friday between 18:00 and 19:00 a regular incident (e.g. a scheduled report) affects the performance of the system. There is no severe slowdowns detection for SAP Integrated Business Planning for Supply Chain.

In summary, the card Event Statistics now provides you with a central location in the SAP EarlyWatch Alert Workspace to evaluate the performance after upgrades and conversions in your system landscape. It displays the most important KPIs for both general system usage and performance, as well as for potential issues regarding specific functions or time slots. The event information is well integrated in the SAP EarlyWatch alert infrastructure offering the possibility of drill-downs in the appropriate views of the SAP EarlyWatch Alert Dashboard.

For more information, see the in-app help of the card and the event list.

[jgleichmann]

Hi dirk.henrich,

thanks for introducing this new feature. I think a lot of customers had such a tooling on there wishlist on their way to SAP HANA or S/4HANA.

May be some more background information are needed to understand how this feature work and what the customers have to do to consume it.

So, what are the prerequisites?

Which statistics are used?

Does it also recognize parameter or hardware changes?

Only ABAP statistic data or even HANA statistic data?

To which level of detail can it be drilled down? (application, user, SQL command, hash etc.)

 

Regards,

Jens

[ajankowiak]

Hello,

 

does event SAP Product Upgrade also mean Service Stack implementation?

 

BR,

Andreas

[SusanneGlaenzer]

Hi Jens,

thanks for the positive feedback and questions! Here are my answers:

• So, what are the prerequisites?

1. Systems have to send EWA data to SAP since 8 weeks prior to upgrade/conversion. ST03 performance collectors have to be scheduled. For details, see SAP note 3110798. You can check that it works by using the EWA dashboard : if the response times statistics are visible, you are good to go.


2. Your S-User needs the authorization "Service Reports & Feedback". If you can access EWA workspace, you are good to go.


3. No additional set-up required. The upgrade/conversion will be detected automatically, if the installation number and system number has no changed!

• Which statistics are used?

Only system-wide ST03 ABAP statistics.

• Does it also recognize parameter or hardware changes?

The EWA service recognizes parameter and hardware changes and documents them in the EWA report which you can find here. However, this is not reflected in the current version of the event statistics app and has to be checked manually in the report.

• Only ABAP statistic data or even HANA statistic data?

The comparison is done with the ABAP statistics only. The corresponding HANA statistics are available in the EWA dashboard and report. They are not compared automatically.

• To which level of detail can it be drilled down? (application, user, SQL command, hash etc.)

In the CPU consumption detail page of the EWA dashboard you can drill down to application source and SQL command. See this blog for details.

 

Best regards,

Susanne Glaenzer

[jgleichmann]

Hi Susanne,

thanks for taking the time and your detailed answer. If the comparison is only based on the ABAP statistic data the HANA native systems can not be used the compare HANA updates/upgrades. May be this part is a little bit missleading inside the blog.

"Well, for the first question: events are significant changes in the technical boundary conditions of a system, such as SAP S/4HANA conversions (which is a migration by upgrading from an existing ERP system to SAP S/4HANA), SAP product upgrades, SAP HANA migrations or SAP HANA upgrades."

But in the end there is no special setup needed for the NW / S/4 systems - just 8 weeks of correctly send data. Well done - it should be easy to use by everyone.

Good job, keep up including more data and features for more scenarios.

May be as an example as feature request:

recognition of NSE activation and compare involved table accesses

Regards,

Jens

[SusanneGlaenzer]

Hello Andreas,

this initial version of the app includes only SAP S/4HANA conversions and product version upgrades, but not Support Packages updates.

In case of ERP, it includes also EHP Upgrades:

e.g. from EHP7 for ERP 6.0 to EHP8 for ERP 6.0

Best regards,

Susanne