C H A P T E R 8 |
System Tab |
The System tab displays performance information about the region, as well as configuration information.
1. Open a region window and select the System tab.
See Displaying a Region's Window.
2. Click the drop-down arrow at the top of the System tab.
3. Select the topic you want to view.
The following topics are available from the System tab:
The Performance Overview window displays information related to overall system overhead, and on a per transaction class basis.
At the top of the pane, there are two datapoints that apply to the region:
The rest of the pane contains datapoints related to the transaction classes defined for the region. You can view the information for individual transaction classes or for all transaction classes. Use the drop-down list to select a transaction class or all transaction classes.
The following datapoints are also displayed:
Total system processor (CPU) time for all transaction processors (in seconds). |
|
Total user processor (CPU) time for all transaction processors (in seconds). |
|
Peak transaction rate per second. See Peak Transaction Rate. |
|
Maximum number of transactions forced to wait for a transaction processor. |
|
Current transaction rate per second. See Transaction Rate. |
|
Total number of transaction executed. See Transactions Executed. |
|
The Performance Details panel displays details related to active transactions.
Three entries at the top of the pane show:
The following datapoints are displayed on this pane:
The Transactions Executed datapoint represents the cumulative total of transactions executed since the region was started.
This value is used to compute the transaction rate for a particular region. As long as your system is operating properly and users are performing transactions, this datapoint should continue to increase.
The Transaction Rate value shows the average number of transactions executed in a second by a region.
The value of this datapoint is computed as follows:
#Transactions executed (since last update) divided by the Polling Interval
This value gives you an idea of a particular region's throughput. This value can be more accurately computed by using a shorter polling interval. However, a shorter polling interval requires substantially more processing overhead.
Peak Transaction Rate shows the highest number of transactions per second (shown with the Transaction Rate datapoint) calculated for a region since connecting to the region.
The datapoint gives you an idea of what your peak processing performance can be. Your system operates with the most throughput when the Transaction Rate datapoint remains consistently near the Peak Transaction Rate.
However, if the Transaction Rate is consistently close to Peak Transaction Rate, it is possible that the region is running at very close to the maximum throughput of the system and additional resources might be required.
Because this value is only calculated when connected to a region, the value may not represent the actual highest transaction rate achieved by the region. That point might have been reached when the region was not being monitored.
The Processes pane of the System tab displays information about the processes running in the region. Any of the following processes can be executing:
For each process, the following information is displayed:
The Logs pane enables you to select and view the following system logs:
Click the radio button next to the log file you want to view.
The log file contents are displayed in the scrollable window.
The Recovery window displays the configuration and status of the recovery file. This window displays the following statistics:
Indicates if recovery is configured. Values are true or false. |
|
Environment variable that identifies the file's location. See Environment. |
|
Indicates if the region uses the standard UNIX file recovery method or the Native Recovery File System (NRFS). See File System. |
|
Size of the recovery file in blocks. See Size. |
|
Last time the (circular) recovery file wrapped. See Last Wrap. |
|
Number of times Sun MTP flushes the file due to a timeout. See Timeouts. |
|
Amount of the file used to store recovery information. See File Usage. |
|
If recovery is active for a region, the Recovery window displays recovery attributes specified in the VSAM Configuration Table (VCT). If recovery is not turned on in the VCT, the Configured datapoint is false and the rest of the datapoints are blank. The minimum required size for a recovery file is 100 blocks. To activate recovery, you must set the size to at least 100 blocks in the VCT and restart the region.
One of the most important datapoints to monitor is the Last Wrap time. This datapoint displays the last time the recovery file became full, usually causing transactions to abort and the database file to roll back. If this condition persists, you should analyze the transactions that are causing this condition. They may not be properly accessing the database or syncpointing. In many cases, batch programs can cause the recovery file to wrap due to syncpointing the database at inappropriate intervals. Another reason for recovery file wrap may be that the recovery file is too small. Increase the size and determine if the Last Wrap value is similar to when the file was smaller.
The Last Wrap datapoint indicates the last time the recovery file became totally full.
If a wrap has not occurred, a zero (0) is displayed. When a wrap occurs, this datapoint displays the date and time of the last wrap.
In Sun MTP, the recovery file is implemented as a circular file. When the file is full, it means records can no longer be written to the file because the tail has caught up with the head of the file. In this case, a transaction must be rolled back to free up the recovery file for further processing. When this happens, the Last Wrap datapoint changes to display the exact time when the wrap occurred. After the rollback occurs, records are again written to the recovery file.
The Last Wrap datapoint must be monitored continuously (and compared to the current time) to catch multiple wraps. When such a condition persists, there are two possible causes:
The File Usage datapoint displays the current amount of space (% of the total file space) required in the recovery file for processing before images. The value fluctuates to reflect the amount of output activity generated by a region.
The File Usage datapoint is a number between 0 and 100, displayed to two decimal places. This datapoint is blank when recovery is not active.
The value of File Usage indicates how a particular recovery file's size is handling the traffic generated by a region. When the value stays close to 100%, and the Last Wrap datapoint continues to change, the recovery file size should be increased.
Many things can affect recovery file usage: application output to disk, deadlocks, looping transactions, program aborts, and so on. If the value continues to indicate the need to increase the size of the recovery file as you actually increase the file's size, examine the application programs for problems. A well-behaved set of applications should have an optimum recovery file size.
The Environment datapoint displays the name of the environment variable that points to the directory where the recovery file is located. The administrator specifies this environment variable in the VCT.
The value of the environment variable can be retrieved on the system running the region using the following command:
where variable-name is the value displayed by the Environment datapoint. Note that this datapoint is blank when recovery is not active.
The value of Environment must be a valid directory path name. The full path name of the recovery file is the value of the Filename datapoint concatenated to the end of the value of the Environment datapoint. The recovery file can be easily moved by changing the value of Environment in the VCT.
The Environment datapoint has no meaning for a native recovery file. For such a file, the Filename datapoint specifies the complete path name of the actual device (the device name is a file in the file system). Refer to the Sun Mainframe Transaction Processing Software Configuration Guide for information about configuring the Native Recovery File System (NRFS).
The Size datapoint indicates the size of the recovery file as configured for the region.
If the size is less than one hundred blocks, Sun MTP does not perform VSAM recovery.
The Size datapoint displays the size of the recovery file in blocks, not bytes. Each block contains the same number of bytes as the VSAM block size for the system. The default VSAM block size is 4 KBytes. However, the block size can be 4, 8, 16, or 32 KBytes. If recovery is not active, the value is zero (0).
The optimum recovery file size depends on the recovery activity generated by applications running within a region. Logically, the recovery file is treated as a circular file. The actual file is a fixed size. In order to safely provide data integrity, Sun MTP must abort a transaction when the logical tail of the file catches up with the head. The value of the Last Wrap datapoint gives an idea of how often this happens. When the Last Wrap datapoint continues to change, recovery file size is a limiting factor in system performance, and the size should be increased.
One way to calculate a working recovery file size is to base the size on the activity of the transaction that performs the most writes. Each write to a database might generate an accompanying before image to the recovery file (if a successive write is to a different record). Multiply the number of these writes by the number of transaction servers. Then multiply that number by the average record size. Use the result as an initial recovery file size (remember to translate the size from bytes to blocks). Adjust the value when the Last Wrap datapoint shows too much activity.
Tip - You must increase the size of the recovery file more if you are using batch jobs that also require recovery file space. |
The Timeouts datapoint displays the number of times Sun MTP flushes before images when an internal (and currently non-configurable) timeout period expires and one of many trigger conditions have not already caused a flush. Sun MTP optimizes recovery file writing by grouping before images together and flushing them all at once. There are a few trigger conditions that cause a flush: reaching a maximum combined size (32 KBytes), reaching a command-line specified number of entries to flush, or experiencing transaction conditions like a syncpoint or end of transaction.
The timeouts datapoint is an integer value initialized to zero at region startup.
The algorithm for writing out before images favors very active systems. You can expect the Timeouts value to be quite low for systems that perform a lot of disk activity. The number of timeouts should increase when a system does not perform very many transactions.
If the Timeouts count increases while the system is quite active, inform your authorized Sun service provider.
This datapoint identifies how I/O to a recovery file is managed. The type of management is determined when the recovery file is configured.
On UNIX systems, a native recovery file might offer superior performance over a normal UNIX file. However, the initial setup is more involved for a native file. First, you must choose a free disk area. This could be a completely free raw partition (one not used by a file system), or an unused portion of a raw partition. The space for a native recovery file must be formatted with the Sun MTP kixnrfs utility. Refer to the Sun Mainframe Transaction Processing Software Configuration Guide for more details. Because it is so easy to set up and resize a UNIX recovery file, you can establish an optimum recovery file size with a UNIX file first, then transfer the file to a native organization.
The Batch panel displays the following information:
Node name of the configured batch system. See Batch System. |
|
Frequency with which the region checks to determine if a batch job is waiting to run |
A region can be configured to connect to a Sun MBM node. This connection allows Sun MBM batch jobs to access the region's VSAM datasets. The Batch System datapoint identifies the node directory, which is configured in the region's VCT.
If a node directory is defined, that value is displayed in this datapoint. Otherwise, the value is N/A.
If the region is not configured, Sun MBM batch jobs cannot access VSAM datasets maintained in that region. A region configured for connection to a Sun MBM node does not guarantee a connection. The connection can be verified by examining system messages written to the region's log file, $KIXSYS/unikixmain.log.
The Languages pane displays the languages used in the programs running in the region.
A box with the language name is displayed in the pane as well as the following datapoints:
Indicates if a language is configured for the region. Values are true and false. |
|
Identifies the directory path where the language is installed. |
|
The Limits panel displays a set of configuration restrictions:
Indicates the maximum number of background tasks configured. |
|
Indicates the number of batch jobs that can run concurrently in the region. |
|
Indicates the number of debug terminals configured. See Maximum Debug Terminals. |
|
Indicates the number of transaction servers configured. See Transaction Processors. |
|
Indicates the number of query jobs that can run concurrently in the region. |
|
The Transaction Processors datapoint displays the number of transaction servers available for executing transactions. Each transaction server is a system process.
The number of Transaction Processors is the number of transaction servers configured for the region.
The Maximum Debug Terminals datapoint indicates how many debug terminals are available for debugging a region.
The value of Maximum Debug Terminals is the number configured for the region.
This datapoint represents the number of transaction servers to be used for executing transactions in debug mode.
If this value is zero, no debugging can be performed.
The Accounting panel displays a set of configuration options.
Copyright © 2004, Sun Microsystems, Inc. All rights reserved.