f8b742db-54c6-45f7-b3ff-28eea5213e5c-160121232551.doc SOA PERFORMANCE METRICS: HOW-TO MONITOR ORACLE ENTERPRISE MANAGER By Krishnaprabha Chari, SOA/Middleware Architect This document details the different performance metrics that are critical to monitoring the overall health of SOA. The first section describes the different parameters that are examined as well as the different sources from where these metrics are extracted. The later sections describe the data collection methodology for the metrics. I. SOA Performance Report Currently, the SOA servers are monitored on a daily basis and a report is generated everyday showing the uptime, response time, CPU usage and other critical metrics. A sample of the report is shown below. Metrics Metric Measures Data Source Baseline (05/29/2009) Response Time Average Response Time Website Pulse 2.0 seconds Up Time FMW uptime (for this month, until today) Website Pulse 96% Scalability # of Request per second Grid Control 5.22/second # of Request - OHS Grid Control 4.8/second SOA Infrastructure Average Connection counts Grid Control 128 CPU usage Grid Control 8.3% Heap Usage(MB) Grid Control 1204 II. Data Collection Methodology for the SOA Report Krishnaprabha Chari Page 1 of 16 01/20/2014
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
This document details the different performance metrics that are critical to monitoring the overall health of SOA.
The first section describes the different parameters that are examined as well as the different sources from where these metrics are extracted. The later sections describe the data collection methodology for the metrics.
I. SOA Performance Report
Currently, the SOA servers are monitored on a daily basis and a report is generated everyday showing the uptime, response time, CPU usage and other critical metrics. A sample of the report is shown below.
Metrics Metric Measures Data SourceBaseline
(05/29/2009)
Response TimeAverage Response Time Website Pulse 2.0 seconds
Up Time
FMW uptime (for this month, until today) Website Pulse 96%
Scalability
# of Request per second Grid Control 5.22/second# of Request - OHS Grid Control 4.8/second
SOA Infrastructure
Average Connection counts Grid Control 128CPU usage Grid Control 8.3%Heap Usage(MB) Grid Control 1204
II. Data Collection Methodology for the SOA Report
Source: Website PulseURL: http://www.websitepulse.com/, login as portals/portalsUnder Reports, go to All Reports/Transactions
Select target: FMW CheckSelect report type: Monthly Log..Select starting date: Start of Month/Date of choice
Generate Report generates the data a shown below.The Average Response Time column shows the average response time on a daily basis. This time is the sum total of the average time taken by all components in FMW Check: Portal, IDM, and SOA.
The best source for getting the Average Response time for BPEL/SOA Processes is the BPEL dehydration database. Run the appropriate queries against the schema (sample query shown in Appendix) to determine this.
2. Computation of Up Time
Source: Website Pulse URL:http://www.websitepulse.com/, login as portals/portalsUnder targets, select FMWCheck
a. Formula for Up Time From the main page of FMW Check, Uptime% shows the % of time the different components of SOA, Portal and EBS were up. As only 1/3 of the downtime is attributable to SOA, we recast the Uptime, only taking 1/3 of the total downtime.
SOA Uptime is calculated as 100- (100 – Uptime% *1/3).
This Month UptimeLocation Total checks Fails Average response time Uptime % DowntimeSeattle 2323 4 1.8395 99.8278 % 0h 18mNew York 2166 3 2.1710 99.8615 % 0h 17m
Total 4489 7 1.9995 99.8441 % 0h 18m
In this example, SOA Uptime would be 100 – (100 – 99.8441 *1/3)
Source: EM GridURL: login with appropriate user credentials.
Scalability is made of two components, # of Requests/sec and # of Requests-OHS.
a. # Of Request per second
Under EM Grid, navigate to Targets/Application Servers TabClick on the SOA Production Node SOA.ora-prod-fmsoa-a1.com (You can also select other nodes from A2 through A4)Choose Component OC4J_SOA and the Performance Tab for this component.
Screen shot below shows the page that comes up. Click on the Requests per second link highlighted below,
On the next page, in the View Data tab, select the time period (as shown 4 pm – 4pm) and then select Compare Targets to select the other nodes (A2-A4). The Compare Targets is a visual check to ensure that the requests are equally distributed across the different nodes. The screen show below shows the resulting page.
i. Formula for Requests per second
Requests/second = Average Value * Number of Nodes (in our current scenario, 4)
Screen shot below shows the page that comes up. Click on the Request Throughput (requests per second) link highlighted below,
On the next page, select prodsoa.com:IP0.0.0.0,Port7777.In the View Data tab, select the time period (as shown 4 pm – 4pm) and then select Compare Targets to select the other nodes (OHS2). The Compare Targets is a visual check to ensure that the requests are equally distributed across the different nodes. The screen show below shows the resulting page.
i. Formula for Requests per second - OHS
Requests/second –OHS = Average Value * Number of Nodes (in our current scenario, 2)
Under EM Grid, navigate to Targets/Application Servers TabClick on the SOA Production Node SOA.ora-prod-fmsoa-a1.com (You can also select other nodes from A2 through A4)Choose Component OC4J_SOA and the Performance Tab for this component.
Screen shot below shows the page that comes up. Click on the Open JDBC Connections link highlighted below,
On the next page, in the View Data tab, select the time period (as shown 4 pm – 4pm) and then select Compare Targets to select the other nodes (A2-A4). The Compare Targets is a visual check to ensure that the requests are equally distributed across the different nodes.
i. Formula for Average Connection Counts Average Connection Counts = Average Value as shown under Statistics
b. CPU Usage/node
Under EM Grid, navigate to Targets/Application Servers TabClick on the SOA Production Node SOA.ora-prod-fmsoa-a1.com (You can also select other nodes from A2 through A4)Choose Component OC4J_SOA and the Performance Tab for this component.
Screen shot below shows the page that comes up. Click on the CPU Usage link highlighted below,
On the next page, in the View Data tab, select the time period (as shown 4 pm – 4pm) and then select Compare Targets to select the other nodes (A2-A4). The Compare Targets is a visual check of the CPU Usage across the different nodes.
The screen show below shows the resulting page.
i. Formula for CPU Usage CPU Usage/node = Average Value as shown under Statistics
Under EM Grid, navigate to Targets/Application Servers TabClick on the SOA Production Node SOA.ora-prod-fmsoa-a1.com (You can also select other nodes from A2 through A4)Choose Component OC4J_SOA and the Performance Tab for this component.
Screen shot below shows the page that comes up. Click on the Heap Usage link highlighted below,
On the next page, in the View Data tab, select the time period (as shown 4 pm – 4pm) and then select Compare Targets to select the other nodes (A2-A4). The Compare Targets is a visual check of the CPU Usage across the different nodes.
The screen show below shows the resulting page.
i. Formula for Heap Usage Heap Usage/node = Average Value as shown under Statistics