Sample Report for Java System

SAP EarlyWatch Alert

The following SAP EarlyWatch Alert (EWA) report for a SAP NetWeaver Portal system gives you an overview of the checks available for the SAP NetWeaver Java stack. It can be used as a reference for the content you can expect to find when you have successfully set up the infrastructure required for the SAP EarlyWatch Alert. For a Java stack system, set up Solution Manager Diagnostics and follow SAP Note 976054. Attached to the note you can find the User Guide for EWA for non-ABAP components. Additional information: For more information about the set up of Solution Manager Diagnostics, see: o http://service.sap.com/instguides Solution Manager Release 4.0 (direct link http://service.sap.com/~form/sapnet?_SHORTKEY=01100035870000673252&) o Here you can find: Wily Introscope Installation Guide Configuration Guide for SAP Solution Manager For more technical information about the SAP EarlyWatch Alerts report content, see the Solution Manager Learning Maps. You can find these documents in SAP Service Marketplace at http://service.sap.com/rkt-solman Solution Manager 4.0 o Select the required Learning Map: Learning Map for Technology Consultants and System Administrator EWA, SL, and Solution Reporting o The following two documents refer to the SAP EarlyWatch Alert for the Java stack: Non-ABAP Components in EWA Overview Non-ABAP Components in EWA Garbage Collection Analysis

Service Report
SAP EarlyWatch Alert

SAP System ID SAP Component Release Status DB System

YYY Enterprise Portal

Not Productive

Customer

Service Center Telephone Fax

Date of Session Date of Report Author

19.11.2007 19.11.2007 EarlyWatch Alert Service

<Watermark>

<dat_dummy>

Session No. Installation No. Customer No.

Service Summary

1 Service Summary
During this EarlyWatch Alert Session, we detected issues that could potentially affect your system. We recommend that you take corrective action as soon as possible.

Note:

All recommendations provided in this report are based on our general experience. We advise that you test our recommendations before using them in your production system. Also note that EarlyWatch Alert is an automatic service.

Section Overview Rating Checked Area Java VM Memory Performance Java VM Activity Reporting Note: For more information about EarlyWatch Alert, a sample EarlyWatch Alert report with explanations is available in the Media Library on SAP Service Marketplace (http://service.sap.com/ewa). This provides an overview of the check rating strategy and the KPIs that trigger the EWA alerts.

Alert Message Overview Note: If you need help investigating the alerts listed, order an EarlyWatch contract by contacting your local support organization or by creating a customer message in component SV-BO. If you already have such a contract, a detailed analysis can be performed during the next Service Session. Address the topic during preparation for the session. If you experience serious performance problems, create a customer message on component SV-BO with the priority "high" or "very high".

Note:

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data

2 Java Performance Data

Reporting Interval
Data displayed in this section of the report was collected for the time frame specified below. System YYY Start Time Sun Nov 11 00:00:00 CET 2007 End Time Sun Nov 18 00:00:00 CET 2007

2.1 Support Package Maintenance

The table below shows the detected version. SUPPORT PACKAGES Version 7.00 PatchLevel 30127.450 JDK VERSION Hostname XXXX601.XXX-group.com Instance VM Vendor ID 8478800 Java HotSpot(TM) 64-Bit Server VM (Sun Microsystems Inc.) VM Version 1.4.2_16b05 SP 13

2.2 Java Memory Configuration

The system is configured with memory parameters in the table below. Host Name XXXX601.XXX-group.com XXXX601.XXX-group.com XXXX601.XXX-group.com Node ID 8478800 8478850 8478851 Node Name dispatcher server0 server1 MaxHeapSize[MB] 170 1024 1024 Xms 170m 1024m 1024m

Background: The "MaxHeapSize" defines the maximum amount of memory that the Java VM provides for Java objects. "Xms" is the minimum amount of memory that is configured.

2.3 Java VM Memory Performance

2.3.1 Garbage Collection Time
Time spent in Garbage Collections (GC) prevents the Java virtual machine from executing any other task. Over the interval reported, this time is not critical.

Average GC Time (Cluster wide) [%] 0,06

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data The table above displays the cluster-wide weekly average of Garbage Collection (GC) time. Garbage collections themselves have a direct impact on system performance. Furthermore, they may also be indicative of more substantial performance issues that may require further analysis. Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478850 8478851 Average [%] 0,06 0,06 Max [%] 0,19 0,17

The table above lists the average and the maximum GC time for the server nodes calculated as defined above. You should investigate particularly high values since these can point to critical situations. As a starting point to analyze errors, you should analyze the log files of the particular server node for the given timeframe. Recommendations: If a high percentage of time is spent in garbage collections, you can measure its impact on performance directly. Thus, if the average value is particularly large, you need to analyze this from a general performance perspective. Temporarily high values of the GC time indicate critical situations. You should analyze these specific situations in comparison with other key performance indicators in this report to determine potential correlations.

2.3.2 Old Space Growth Rate

Critical memory situations in Java are often characterized by a sudden growth of the memory used. Rapid growth of memory used on old space is the most critical, since this area is required for longer living objects. A rapid growth rate quickly leads to a resource outage, and so high growth rates should be investigated further. The growth rate in the old space is calculated using a sliding window approach, in which the last 10 full garbage collections are analyzed. A linear trend analysis is performed on this interval. This approach is chosen so that short intervals of rapid growth in the old space can be detected. This kind of growth is typical of out of memory situations. The maximum values determined during the reporting interval for each server node are reported here. Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478851 8478850 Max [KB/s] 621.21 596.50

Thresholds: Small, non-negative growth rates are not critical and are within the statistical fluctuations usually observed for this parameter. To assess the potential criticality, you have to compare this number to the available heap size, and in particular, the size of the old space. Recommendation: Further analysis is required to determine the cause of rapid memory growth phases. In many cases, these are accompanied by increases in load, which in turn are often characterized by increases in HTTP sessions. The first stage is to analyze these correlations.

2.3.3 Usage of Effective Old Space

The resource utilization of the effective old space is not critical. The higher the value, the less leeway there is for high load situations.

Usage of Effective Old Space (Cluster wide)[%] 17,73 The effective old space is defined as the part of the old space that triggers a full garbage collection if filled completely. In a simple definition, it can be regarded as the old space minus the new space, since a full garbage collection is triggered if the old space does not have sufficient space to receive the new space.

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478850 8478851 Average [%] 18,05 17,40

2.3.4 Fraction of Full Garbage Collections

The maximum ratio of full garbage collections to all garbage collectionsin the reported interval was higher than 50%. That means from 10 consecutive garbage collections were more than 5 full garbage collections. Such phases could indicate critical situations. The cause of this should be investigated further by comparing the other parameters calculated in this section.

Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com

Cluster Element ID 8478850 8478851

Max [%] 80.00 80.00

In general, minor garbage collections must occur more frequently than full garbage collections. A consecutive number of full garbage collections is a clear indicator that the Java Virtual Machine (JVM) is not removing objects from the old space. If the dynamic behavior of the JVM is dominated by old space memory removal, insufficient resources are available to accept requests and their newly-created objects. In this case, the parameter below is analyzed. On a floating interval of all garbage collections, the fraction of full garbage collections is calculated.

2.3.5 Allocation Rate

The allocation rate quantifies the amount of memory allocated on the young generation each second. This therefore quantifies the input to the Java VM. Particularly high values indicate high JVM throughput and so should always be analyzed in parallel with the user load or the number of HTTP sessions. Allocation Rate (Cluster wide)[KB/s] 1604 Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478851 8478850 Average [KB/s] 1888 1320

2.3.6 Promotion Rate

The promotion rate quantifies the amount of memory transferred to the old space each second. Due to the infant mortality of Java objects, this value should always be much smaller than the allocation rate. One indicator of critical situations is a high ratio of promotion rate to allocation rate, since this implies efficient use of the Eden space cannot be made. Promotion Rate (Cluster wide)[KB/s] 14 Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478850 8478851 Average [KB/s] 14 13

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data

2.4 Java VM Activity Reporting

This chapter displays the distribution of activity per server node. If multiple instances are configured, the values give an overview of the balanced load provided that all nodes were running during the reporting interval.

2.4.1 HTTP Sessions

Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478850 8478851 Average 1.00 1.00 Max 4.00 8.00

The above table displays the weekly average and maximum value of HTTP sessions for each server node. A large difference in the values between different server nodes could be caused by unbalanced load distribution.

2.4.2 Active Application Threads

Hostname XXXX601.XXX-group.com XXXX601.XXX-group.com Cluster Element ID 8478850 8478851 Average 1.00 1.00 Max 3.00 4.00

The above table displays the weekly average and maximum value of active Application Threads per server node. A high difference between the values for different server nodes could be caused by unbalanced load distribution or different configuration.

2.5 Java Application Performance

2.5.1 Top WebDynpro Applications
This section measures and reports the average response time over the reporting interval and the accumulated time of the identified WebDynpro application.

2.5.1.1 Response Time

The following table lists the top Web Dynpro applications sorted by the average response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank Name 1 2 3 4 local/WDKIBCustomerData/WDKIBCustomerDataApp local/WDKIBMUWData/WDKIBMUWDataApp local/WDKIBKKDData/WDKIBKKDDataApp local/WDCustomerPDF/WDCustomerPDFApp Average Response Time [ms] 2899 1901 1311 830 Executions 92 73 91 17

Recommendations: Investigate the tuning potential of Web Dynpro applications that have persistently high average response times.

2.5.1.2 Accumulated Time

The following list is sorted by accumulated response time during the reporting interval. The list can be used to determine which application effectively caused the greatest amount of load on the J2EE, since it is not only influenced by the average response time but also by the number of instances the application created at runtime. Rank 1 2 3 Name local/WDKIBCustomerData/WDKIBCustomerDataApp local/WDKIBMUWData/WDKIBMUWDataApp local/WDKIBKKDData/WDKIBKKDDataApp Accumulated Time [h] 0.07 0.04 0.03 Executions 92 73 91 6

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data Rank 4 Name local/WDCustomerPDF/WDCustomerPDFApp Accumulated Time [h] 0.00 Executions 17

Recommendations: Web Dynpro applications with a large accumulated time potentially have a larger overall performance impact. This may be worth further investigation.

2.5.2 Top Servlets

Both the average response time over the reporting interval as well as the accumulated time of the identified servlet is measured and reported here.

2.5.2.1 Response Time

The following table lists the top servlet/JSP requests sorted by the average response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank 1 2 3 4 5 6 7 8 Name SPMLConsumer UserAdminServlet DispatcherServlet JSPServlet Gateway Dispatcher DataSupplierServlet CIMOMServlet Average Response Time [ms] 9164 3533 2748 2665 1702 1158 225 137 Executions 9 1 288 2 90 827 10 42

Recommendations: The performance tuning potential of servlet/JSP requests with high average response times should be investigated.

2.5.2.2 Accumulated Time

The following top list is sorted by the accumulated response time during the reporting interval. This is useful for determining which application placed the greatest load on the J2EE, as this is influenced by both the average response time and the number of times that the application is created at runtime. Rank 1 2 3 4 5 6 7 8 Name Dispatcher DispatcherServlet Gateway SPMLConsumer CIMOMServlet JSPServlet UserAdminServlet DataSupplierServlet Accumulated Time [h] 0.27 0.22 0.04 0.02 0.00 0.00 0.00 0.00 Executions 827 288 90 9 42 2 1 10

Recommendations: Servlet/JSP requests with a high accumulated time potentially have a larger overall impact on performance that may be worth investigating further.

2.5.3 Top JCO Calls

2.5.3.1 Response Time
The following table lists the top JCo calls sorted by the average response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster.

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data Rank Name 1 2 3 4 5 6 7 8 9 10 MSP101@XXXX402sap|JCO|/XXX/RFC_CS_IFLO_HIERARCHY2 MSP-101@XXXX422|JCO|RFC_GET_FUNCTION_INTERFACE MSP-101@XXXX423|JCO|/XXX/RFC_CUSTOMER_MUW MSP-101@XXXX424|JCO|DD_GET_FIELD_INFO MSP-101@XXXX423|JCO|/XXX/PORTAL_GET_DOCUMENTS MSP-101@XXXX422|JCO|DD_GET_FIELD_INFO MSP-101@XXXX423|JCO|DD_GET_FIELD_INFO MSP-101@XXXX421|JCO|DD_GET_CURRENCIES MSP-101@XXXX421|JCO|DD_GET_FIELD_INFO MSP-101@XXXX424|JCO|/XXX/RFC_CUSTOMER_MUW Average Response Time [ms] 4364 1385 987 860 828 726 514 484 403 301 Executions 13 14 4 5 2 5 13 4 18 3

Recommendations: JCo calls with high average response times should be investigated with respect to the performance of the associated back-end application or, if all JCo calls show persistently bad performance in general, with respect to potential network-related performance issues. The following top list is sorted by the accumulated response time during the reporting interval. Rank Name 1 2 3 4 5 6 7 8 9 10 MSP101@XXXX402sap|JCO|/XXX/RFC_CS_IFLO_HIERARCHY2 MSP-101@XXXX422|JCO|RFC_GET_FUNCTION_INTERFACE MSP-101@XXXX421|JCO|DD_DOMA_GET MSP-101@XXXX421|JCO|RFC_GET_FUNCTION_INTERFACE MSP-101@XXXX421|JCO|DD_GET_FIELD_INFO MSP-101@XXXX423|JCO|DD_GET_FIELD_INFO MSP-101@XXXX423|JCO|DD_DOMA_GET MSP-101@XXXX424|JCO|DD_GET_FIELD_INFO MSP-101@XXXX423|JCO|/XXX/RFC_CUSTOMER_MUW MSP-101@XXXX422|JCO|DD_GET_FIELD_INFO Accumulated Time [h] 0.02 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Executions 13 14 606 39 18 13 406 5 4 5

Recommendations: Investigate the potential overall performance impact of JCo calls with high accumulated times.

2.5.4 Top SQL Selects

This section analyzes the top SQL select statements observed. These statements are presented as top lists and are sorted by both average response time and accumulated time.

2.5.4.1 Response Time

The following table lists the top SQL select statements sorted by the average response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank Name Average Response Time [ms] 1814 1173 330 165 Executions

1 2 3 4

YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_ATTR_VALUES" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_ATTR_HEADERS" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_OBJECTS" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "FILE_CONTENTS" FROM "MM_CS_EXT_FILES" WHERE "FILE_NAME" = ?

10 10 10 20

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data Rank Name Average Response Time [ms] 133 Executions

6 7 8 9

10

YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "SESSIONUID", "PRINCIPAL", "APPLICATION", "CREATIONTIME", "CLIENTCOUNT", "MM_STATE", "STATETIME", "USESSUBSCRIBER", "PRODUCTINFO1", "PRODUCTINFO2", "PRODUCTINFO3", "PRODUCTINFO4" FROM "MM_SERVICESESSIONS" WHERE "MM_STATE" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_SHORT_STRS" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT DISTINCT "DSTID" FROM "BC_JMSTOPIC" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT MAX ("TIMESTMP") FROM "BC_SYNCLOG" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|UPDATE "J2EE_CONFIGENTRY" SET "FBLOB" = ? WHERE "CID" = ? AND "NAMEHASH" = ? AND "NAME" = ? AND "ISFILE" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_ATTRS1"

10

109 78 52 50

10 10 10 675

36

10

Recommendations: Investigate the performance of the database associated with SQL select statements that have high average response times. If all statements show persistently bad performance in general, investigate potential networkrelated performance issues.

2.5.4.2 Accumulated Time

The following top list is sorted by the accumulated response time during the reporting interval. Rank Name 1 YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "CONTENTID", "INTCONTENTLENGTH" FROM "KMC_DBRM_NODES" WHERE "ID" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|UPDATE "J2EE_CONFIGENTRY" SET "FBLOB" = ? WHERE "CID" = ? AND "NAMEHASH" = ? AND "NAME" = ? AND "ISFILE" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "CPATH", "CID", "PARENTCID", "CACHEMODE", "CTYPE" FROM "J2EE_CONFIG" WHERE "PATHHASH" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "ID", "ATTR_NAME_NUM", "DELTA_LINK_TYPE", "SUBTREE_ROOT_ID", "ORDERED", "ATTR_VAL_TYPE", "INHERITANCE_TYPE", "STORAGE_TYPE", "WRITE_PERM_TYPE", "VALUE_COUNT", "VALUE_LENGTH", "INT_VALUE", "DOUBLE_VALUE", "STR_SHORT_VALUE", "TEXT_TYPE", "LOCALE_LANGUAGE", "LOCALE_COUNTRY", "STR_LARGE_VALUE" FROM "EP_ATTR_HEADERS" WHERE "ID" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "ID", "ATTR_NAME_NUM", "DELTA_LINK_TYPE", "SORT_INDEX", "SUBTREE_ROOT_ID", "VALUE_LENGTH", "INT_VALUE", "DOUBLE_VALUE", "STR_SHORT_VALUE", "LOCALE_LANGUAGE", "LOCALE_COUNTRY", "STR_LARGE_VALUE" FROM "EP_ATTR_VALUES" WHERE "SUBTREE_ROOT_ID" = ? ORDER BY ?, ?, ?, ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT (*) FROM "EP_ATTR_VALUES" YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|Unknown SQL YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "CONTENTID", "MODIFYCOUNTER", "INTCONTENTLENGTH", "INTCREATIONDATE", "INTLASTMODIFIED" FROM "KMC_DBRM_NODES" WHERE "ID" = ? Accumulated Time [h] 0.03 Executions 22570

0.01

675

0.01

19992

0.01

10534

0.01

13234

6 7 8

0.01 0.00 0.00

10 16737 953978

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

Java Performance Data Rank Name 9 Accumulated Time [h] Executions 23734

10

YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT "ID", 0.00 "SUBTREE_ROOT_ID" FROM "EP_OBJECTS" WHERE "PARENT_ID" = ? AND "NAME" = ? YYY XXXX601-1527 (Oracle DB)|SQL|Prepared|Query|SELECT COUNT 0.00 (*) FROM "EP_ATTR_HEADERS"

10

Investigate the potential overall performance impact of SQL statements with high accumulated times.

2.6 Enterprise Portal Activity Reporting

2.6.1 Hourly User Distribution
The following diagram displays the hourly user distribution calculated by stacking each day and constructing hourly averages. Note that the time zone displayed refers to that configured for shared master data. A distinction is made between anonymous users and named users. Note that this presentation spans the complete cluster.

Hourly User Distribution

12 10 8 6 4 2 0 4 5 6 7 8 00 01 02 03 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Anonymous Named

Recommendations: This diagram allows you to determine the hour of the expected peak load. If the critical situations detected are caused by user load, these events should correlate with the user distribution measured.

2.6.2 Top Iviews

2.6.2.1 Response Time
The following table lists the top iViews sorted by the average response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank Name Average Executions Response Time [ms] 2928 2168 2154 2080 646 22 19 45 19 12

1 2 3 4 5

com.sap.portal.navigation.portallauncher.default|<UnknownRole> com.sap.portal.pageToolbar|pcd-portal_content/every_user/general/defaultDesktop/ frameworkPages/frameworkpage com.XXX.KIBDepotWD_20060609_rel2|pcdportal_content/XXX/KIB_WD/WDKIB/WDKIBPage _rel01 com.sap.portal.masthead|pcd-portal_content/every_user/general/defaultDesktop/fra meworkPages/frameworkpage com.sap.portal.contentCatalogTreeContent|pcd-portal_content/administrator/super_ admin/super_admin_role/com.sap.portal.content_administration/com.sap.portal.cont ent_admin_ws/com.sap.portal.portal_content com.sap.portal.contentCatalogSearchContent|pcdportal_content/administrator/supe r_admin/super_admin_role/com.sap.portal.content_administration/com.sap.portal.co ntent_admin_ws/com.sap.portal.portal_content com.sap.portal.pcmEditor|pcd-portal_content/com.sap.pct/admin.templates/iviews/e ditors

395

362

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

10

Java Performance Data Rank Name Average Executions Response Time [ms] 6 12 1

8 9 10

com.sap.portal.exportWizard|pcd-portal_content/com.sap.pct/admin.templates/iview 314 s/editors com.sap.portal.export|pcd-portal_content/com.sap.pct/admin.templates/iviews/edit 301 ors Midsys-L03|pcd-portal_content/XXX/XXX_ROLLEN/Midsys-L03 249

Recommendation: Investigate the performance tuning-potential of iViews with persistently high average response times. Note, however, that this parameter alone does not take into account the frequency with which the iView was called.

2.6.2.2 Accumulated Time

The following table lists the top iViews sorted by the accumulated response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank Name 1 2 3 4 5 6 7 com.XXX.KIBDepotWD_20060609_rel2|pcdportal_content/XXX/KIB_WD/WDKIB/WDKIBPage _rel01 com.sap.portal.navigation.portallauncher.default|<UnknownRole> com.sap.portal.pageToolbar|pcd-portal_content/every_user/general/defaultDesktop/ frameworkPages/frameworkpage com.sap.portal.masthead|pcd-portal_content/every_user/general/defaultDesktop/fra meworkPages/frameworkpage com.sap.portal.runtime.logon.default|<UnknownRole> com.sap.portal.runtime.logon.certlogon|<UnknownRole> com.sap.portal.contentCatalogTreeContent|pcd-portal_content/administrator/super_ admin/super_admin_role/com.sap.portal.content_administration/com.sap.portal.cont ent_admin_ws/com.sap.portal.portal_content com.sap.portal.contentarea|pcd-portal_content/every_user/general/defaultDesktop/ frameworkPages/frameworkpage/com.sap.portal.innerpage com.sap.portal.layoutPortal1_WAandNavPanel|pcdportal_content/every_user/general /defaultDesktop/frameworkPages/frameworkpage/com.sap.portal.innerpage com.sap.portal.export|pcd-portal_content/com.sap.pct/admin.templates/iviews/edit ors Accumulated Executions Time [h] 0.03 0.02 0.01 0.01 0.00 0.00 0.00 45 22 19 19 86 43 12

8 9

0.00 0.00

87 87

10

0.00

12

Recommendation: Investigate the performance tuning-potential of iViews with persistently high accumulated response times. Note that this parameter takes into account the frequency with which the iView was called. Therefore, the tuning performance of an iView with high accumulated time will have a larger overall effect.

2.6.3 Top Components

SAP Portal Component runtime performance is identified by the response time of the doContent method. The average response time and the accumulated time are displayed here as top lists.

2.6.3.1 Accumulated Time

The following table lists the top components sorted by the accumulated response time. Note that the average was calculated across all application nodes of the SAP J2EE cluster. Rank Name 1 com.XXX.KIBDepotWD_20060609_rel2|pcdportal_content/XXX/KIB_WD/WDKIB/WDKIBPage _rel01 Accumulated Executions Time [h] 0.03 45

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

11

Java Performance Data Rank Name 2 3 4 5 6 7 com.sap.portal.navigation.portallauncher.default|<UnknownRole> com.sap.portal.pageToolbar|pcd-portal_content/every_user/general/defaultDesktop/ frameworkPages/frameworkpage com.sap.portal.masthead|pcd-portal_content/every_user/general/defaultDesktop/fra meworkPages/frameworkpage com.sap.portal.runtime.logon.default|<UnknownRole> com.sap.portal.runtime.logon.certlogon|<UnknownRole> com.sap.portal.contentCatalogTreeContent|pcd-portal_content/administrator/super_ admin/super_admin_role/com.sap.portal.content_administration/com.sap.portal.cont ent_admin_ws/com.sap.portal.portal_content com.sap.portal.contentarea|pcd-portal_content/every_user/general/defaultDesktop/ frameworkPages/frameworkpage/com.sap.portal.innerpage com.sap.portal.layoutPortal1_WAandNavPanel|pcdportal_content/every_user/general /defaultDesktop/frameworkPages/frameworkpage/com.sap.portal.innerpage com.sap.portal.export|pcd-portal_content/com.sap.pct/admin.templates/iviews/edit ors Accumulated Executions Time [h] 0.02 0.01 0.01 0.00 0.00 0.00 22 19 19 86 43 12

8 9

0.00 0.00

87 87

10

0.00

12

Recommendation: Investigate the performance tuning-potential of components with persistently high accumulated response times. Note that this parameter takes into account the frequency with which the component was called. Therefore, the tuning performance of a component with high accumulated time will have a larger overall effect.

EarlyWatch Alert Report, Sample Report on SAP Service Marketplace, 19.11.2007

12