70 WebSphere Studio Application Monitor V3.2 Advanced Usage Guide
Load burst: A sudden additional load followed by an increase of response
time. If the increase is more than proportional to the load increase, then it may
be an application problem that is not tuned properly to handle more load.
Some special situations may occur for new or changed applications. These
changes require the operator to take on additional monitoring tasks. This
includes performance tuning for the new application, as discussed in 4.4,
“Performance tuning” on page 78. Also, additional reports may be generated to
understand the impact of the new or changed application in the production
WebSphere Studio Application Monitor offers an extensive set of reports about
collected data. These are accessible via PERFORMANCE ANALYSIS →
CREATE REPORTS. The data in these reports is from the database and offers
the best look at historical data.
Reports are useful for:
Providing summary information about application server performance in an
interval. This can also be scheduled regularly with PDF output.
Showing long-term trends for growth of resource usage to allow preventative
maintenance to occur.
Assisting in capacity predictions of potential bottlenecks to allow for
maintenance upgrades or preparation of new hardware.
Comparing information from different time intervals, such as having a
baseline for a good day to compare with a bad day.
4.3.1 Report types
This section contains descriptions of the report types that are available with
WebSphere Studio Application Monitor: application reports and server reports.
Six reports are designed to monitor applications:
Request/Transaction Compares requests over time and drills into specific
requests and their transaction breakdown/flow. Useful for
finding past errant behavior and examining the causes
by going into detail using the drill-down feature.
Chapter 4. \Managing production environment 71
Similar to the Request/Transaction report but based on
methods instead of requests to cover more data. Best
used with the Request/Transaction name or
Method/Program filters to narrow the amount of data
SQL Determines the frequency and response times of SQL
statements. Best used to determine either slowest or
most frequent SQL statements to investigate
performance of specific SQL statements using the
filtering (for example, response times of UPDATE calls to
a specific table).
MQI Similar to the SQL report in that it determines slow or
frequent MQ statements either through filtering or
broadly without filtering.
Lock Analysis Lock data is collected only if Lock Analysis is turned on.
(See 5.3, “Deadlock management” on page 112.) Lock
Analysis reports help identify the frequency, source, and
duration of locks.
Top Reports Basic table reports for finding slowest methods, slowest
SQL call, and so on. No further drill-down is available,
but the aggregate “All Servers from a specific group”
option is useful to quickly see from a high level what is
slow or frequently called across the monitored
Chapter 4. \Managing production environment 73
Typically the application report is presented in a trend report with bar charts, and
you can drill down into a decomposition report that provides a breakdown of each
bar entry into a pie chart by transaction type or transaction name. Figure 4-2
shows the decomposition chart by transaction name.
Figure 4-2 Request report decomposition
74 WebSphere Studio Application Monitor V3.2 Advanced Usage Guide
Drilling down from the decomposition, you can see additional detail, such as
method trace, transaction stack, and SQL call details. Figure 4-3 shows
transaction detail for a specific transaction.
Figure 4-3 Request report details after drilling down
WebSphere Studio Application Monitor has three server reports to monitor
overall server information:
System Resource Reports on overall system resources such as CPU and
memory with no drilldown offered. Good for judging
memory growth, active sessions, and so on.
Server Availability Useful for SLAs, this report is based on every minute
check of the server availability that WebSphere Studio
Application Monitor does.
Capacity Analysis Analyzes information and generates usage trends. This
report plots either throughput per minute or number of
users on the X-axis against system CPU, JVM/process
CPU, JVM/process memory, thread pool, connection pool,
or response time on the Y-axis.