This monitor shows sample or cumulative data, depending on which statistic type you choose.
The graphs on this monitor show sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
If you increase maximum network packet size, check that the additional network memory configuration parameter is set large enough to accommodate requests for larger packet sizes. If there is not enough additional network memory configured to handle a large packet request, the connection receives network I/O buffers equal to the default network packet size even though it logged on requesting a larger size.
The table shows sample or cumulative data, depending on which statistic type you choose.
The graphs on this monitor show sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
The device I/O counts the number of reads to and writes from the buffer. The number of data pages included in one device read or write depends on the buffer size. For a 2K buffer pool (the default size if you are not using large I/O), one device I/O handles one 2K data page. For an 8K buffer pool, one device I/O handles four 2K data pages.
The data page I/O counts the number of data pages, regardless of the buffer size.
The graphs on this monitor show sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
This monitor shows sample data only. The information is a snapshot of the lock status at the end of the most recent sample interval.
For a selected process, this monitor shows information about the batch text or stored procedure that was executing at the time of the most recent sample. The information refreshes at each sample interval. If a batch or stored procedure starts after one sample and ends before the next sample, its text will not be displayed in this monitor.
For each client connection, text of only the currently executing SQL batch is stored. The text of a new batch overwrites the text of previously executed batches. If the batch text is larger than the value of max SQL text monitored, only the beginning of the batch text is stored. The remainder is truncated.
The graph shows sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
This monitor shows sample or cumulative data, depending on which statistic type you choose.
The graphs on this monitor show sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
The graphs on this monitor show sample or cumulative data, depending on which statistic type you choose. The table portion is not affected by the statistic type. The table portion shows:
Object Page I/O Monitor Details
Network Activity Monitor Details
The following Adaptive Server configuration parameters are useful in evaluating network performance.
On busy networks, the packet transfer rates can also depend on the amount of other traffic not related to Adaptive Server and the packet per second capacity of the network hardware.
On busy networks, traffic not related to Adaptive Server and the packet-per-second capacity of the network hardware has an impact on these rates.
If this value is close to the default network packet size, it means that on average, packets are full. It could mean that the packet size is not large enough to handle average requests, and, as a result, many requests are being divided into multiple packets. This situation can affect Adaptive Server task-level performance. When a CPU task receives a packet in a multi-packet batch or command, it sleeps while waiting for the next packet in the batch or command.
If this value is close to the default network packet size, it means that on average, packets are full. It could mean that the packet size is not large enough to handle average requests, and, as a result, many requests are being sent in multiple packets. This situation can affect Adaptive Server task-level performance. Immediately after a CPU task sends a packet, it sleeps. A task can sleep many times if it is sending a large amount of data in small packets.
Connections that send or receive large amounts of data across the network can achieve significant performance improvement by using larger packet sizes. Generally, you want to keep the value of default network packet size small for users performing short queries, and allow users who send or receive large volumes of data to request larger packet sizes when they log on.
This parameter affects the amount of memory, in bytes, reserved at server boot time for network I/O buffers.
Process Activity Monitor Details
Notes
The Process Activity Monitor handles the volatile status of processes as follows:
Application Activity Monitor Details
In some cases, raising the execution class of a poorly performing application can make performance worse. It is important to evaluate how this applicationÆs execution attributes interact with those of other concurrently running procedures, logins, and applications. To see which applications, stored procedures, and logins are currently bound to an execution class, open the Bindings tab on the execution class's property sheet.
Device I/O Monitor Details
Since requests can be made in one sample interval and completed in another, the rate might be skewed slightly downward. For the same reason, more requests could be granted than made during a sample interval, causing the rate to be greater than 100%.
High (but not much over 100%) rates are desirable. Low rates indicate that requests are not being granted. Values consistently over 100% indicate significant delays in granting requests.
The higher the value, the more read activity on the device. Compare rates for different devices to help in analyzing load balances across devices. High values that match or are approaching the physical limitation of the I/O device are undesirable, indicating that Adaptive Server might wait for I/O.
The higher the value, the more write activity on the device. Compare rates for different devices to help in analyzing load balances across devices. High values that match or are approaching the physical limitation of the I/O device are undesirable, indicating that Adaptive Server might wait for I/O.
Device I/O Counts versus Page I/O Counts
Device I/O counts are not the same as page I/O counts.
Data Cache Monitor Details
This metric indicates relative efficiency of space allocations among data caches.
High values are desirable. The higher the value, the more efficient the space usage. Very high values mean that space is appropriately allocated to the named cache.
Very low values (low hit rates and large amounts of memory) could mean that space is being wasted. Redistributing that space (to the default cache, for example) might improve the Hit Percent of another named cache.
High values are desirable, indicating that a high number of prefetched pages in large I/O grabs are actually being used. Low values indicate that large I/O in this data cache is not providing much benefit. Low values could also mean that a large table is fragmented enough to render large I/O ineffective.
Note:Regardless of how many buffer pools are configured in a named data cache, Adaptive Server only uses two of them. It uses the 2K buffer pool and the pool configured with the largest-sized buffers. Prefetch effectiveness does not apply to the 2K buffer pool, since 2K grabs are not considered large I/O Therefore, prefetch effectiveness applies to the largest buffer pool in the cache. For example, if a data cache has pools of buffers sized 2K, 8K, and 16K, the 8K pool is not used, and this metric reflects the effectiveness of large I/O in the 16K buffer pool.
High hit percents are desirable, since they indicate less time spent waiting on device I/O. Low hit rates indicate that the named data cache size is too small for the current load.
Very large tables with random page access generally show a low cache hit ratio.
High rates are desirable because it indicates more pages satisfied from cache.
Low rates are desirable because physical I/O is time-intensive. High values are normal at the beginning stages of use (for example, when Adaptive Server or an application starts) because data caches start empty and fill up during use.
Low rates are desirable. High rates indicate that the Adaptive Server optimizer wants to perform large I/O and cannot (because a buffer is not available or the data is not contiguous).
Object Lock Status Monitor Details
Low values are desirable. High values indicate that many processes are waiting in line for a lock on this page or table.
Process Current SQL Statement Monitor Details
Select the SPID whose currently executing SQL text and query plan you want to see.
When STP Line Number is present, Batch Line Number is blank.
When stored procedure text is present, the Batch Statement Number is blank.
The amount of batch text available for viewing is controlled by the Adaptive Server configuration parameter max SQL text monitored. If this parameter is zero or too small, batch text is blank or incomplete. Changing this parameter requires rebooting Adaptive Server.
The STP information is blank if:
Rarely, query plan could be blank if the plan was removed from Adaptive ServerÆs cache of query plans before Monitor Server could access it.
Notes
If no information appears in the text portions of this monitor, some reasons might be:
This monitor is not a trace tool. It displays the currently executing statement at the time the sample is taken, and many short-running statements can be missed between samples. Use the Process Activity Monitor to see accumulated resource usage for all statements executed by a process. If you need to watch every SQL statement as it executes, use a debug or trace tool.
To set max SQL text monitored
The max SQL text monitored configuration parameter controls whether SQL batch text is collected and how much text is stored per client connection. The parameter specifies the number of bytes reserved per user connection in the shared memory area shared by Adaptive Server and Monitor Server. If the value of max SQL text monitored is 0, no batch text is stored.
Transaction Activity Monitor Details
The transaction count indicates total transaction throughput handled by Adaptive Server. In systems where transaction activity level is predictable, you can compare benchmarked values to current activity to gauge the health of the system.
Multidatabase transactions are counted per database affected. Rollbacks are not counted. Transactions from all applications are counted, including Adaptive Server Monitor applications using Monitor Client Library. If a transaction starts in one sample interval and completes in the following sample interval, it is counted in the second sample interval only.
Low values are desirable.
Low values are desirable. These updates increase lock contention and page I/O, and generally take more time to complete.
In general, lower values are preferable. High insert activity on heap tables can create lock contention and I/O bottlenecks.
Stored Procedure Activity Monitor Details
In some cases, raising the execution class of a poorly performing stored procedure can make performance worse. It is important to evaluate how this procedureÆs execution attributes interact with those of other concurrently running procedures, logins, and applications. To see which applications, stored procedures, and logins are currently bound to an execution class, click on the Bindings tab on the execution class's Property Sheet.
Engine Activity Monitor Details
Values lower than 60 percent are desirable. Values approaching 60 percent indicate that the engine might be overloaded some of the time. If all engines are over 60%, adding another engine might improve performance.
Some engines show no activity if there are fewer processes than CPUs.
Cache Monitor Details
The higher the value, the more activity in Adaptive Server.
Low rates are desirable because physical I/O is time-intensive. High values are normal at the beginning stages of use (for example, when Adaptive Server or an application starts) because data caches start empty and fill up during use.
High hit rates are desirable, since they indicate less time spent waiting on device I/O. Low hit rates indicate that the cache size is too small for the current load.