Performance collects performance data from targeted monitoring servers and stores it for a period of time. Even after the lapse of time, you can check the performance data of past time and analyze the cause and take action.
The collected data is stored for seven days, and can be adjusted by the user’s physical disk size.
Select a target server
- Servers can only be selected as a single serve, multiple selections are not allowed. It is configured to select one monitored server and analyze it in detail.
To view performance data for a specific date
- To view performance data for a specific date, select a screen calendar to specify a date. For today and yesterday's frequently used dates, the "Today" and "Yesterday" buttons allow easy and quick transitions.
- There is a button on the left side of the date selection that allows you to open and close the time zone selection. The default is that the time zone selection is open.
To set the monitoring time zone.
- The Timeline bar at the bottom of the server and date selection shows data for the latest three hours based on the current time. You can easily view data from the time zone you want by clicking the time in the Time line bar.
- Use the Time-Sync feature to share the time you are currently looking at when you switch between screens. The shared area is the date selected for the calendar and the time zone you select for Timeline bar.
- On a one-minute monitoring screen, you can drag the chart and select the time in minutes. To the right of Time-Sync, the selected time in minutes is displayed.
- Graph of the frequency of performance by type of wait class for the collected session. Use this graph to check the frequency of occurrence by the nature of Wait Class.
- This is a one-minute section graph.
- Displays the top SQL of the server's time zone by interval.
- We are currently displaying the two top SQL in the order of frequency.
- Activity : Run Count basis
- Buffer Reads : The formula is (buffer_gets / executions) that shows the percentage of the average buffer usage when SQL runs.
- When you click on the sql_id, the SQL Detail screen is displayed.
- This is a 2nd touch screen where you can check various performance indicators of the selected SQL. In addition to SQL Text, you can see Plan information, Bind value, and various performance indicators in the last five minutes.
- At the top, there is an area where you can check the server and SQL ID, and you can easily check the SQL information at the desired time through the date selection function.
- The SQL Text area and Plan area can be viewed larger using the right side extension button on each header section.
- You can view the SQL syntax in the SQL Text area, and it provides the ability to view the original text, view the sorting text, and copy professionally. The Sql text(bind) tab automatically inserts bind variables into the SQL syntax. It provides additional SQL syntax correction, real-time plan loading.
- Provides a chart that allows tracking in minutes in the Sql Tracking area. Click the chart area to activate the one-minute move button. The monitoring includes the following:
- disk_reads : The DELTA value of the previous value in minutes.
- buffer_gets : The DELTA value of the previous value in minutes.
- user_io_wait_time : The DELTA value in minutes prior to this value, in ms.
- cpu_time : DELTA of the previous value in minutes, in ms.
- elapsed_time : elapsed_time divided by the execution and the DELTA value of the previous value in minutes. The unit is ms.
- Unit Average Performance Time"
- Executions : DELTA value of the previous value in minutes.
- The Sql info area displays detailed sql information for the selected time zone in SQL Tracking.
- The Plan tab shows the PLAN information about the sql_id in the selected time zone in SQL Tracking. If you have more than one child_number, you can select and view it.
- The bind tab shows the bind capture information for the nearest time zone to the sql_id. It's actually collected every 15 minutes, so it’s at least 15 minutes of data.
- Displays the top sessions for each time zone on the server. Displays the highest top section as a percentage based on sid, serial, and sql_id.
- CPU Used : Converts the value of "CPU Used by this session" for each session to the delta value for each collection period, and displays the CPU-intensive sessions in high order by percentage.
- Execute Count : Displays the Execute Count for the Session in descending order by percentage.
- Logical Reads : Converts Session Logical Reads into the Delta value for the collection cycle and shows the percentage of session with a higher value.
- Physical Reads : Converts Physical Reads to the Delta values for the collection cycle and shows the percentage of session with a high value.
- Redo : Shows the percentage of session with a higher value by converting the session redosize value to the Delta value of the collection period.
- The Stat chart consists of tab divided by each indicator characteristic. If you want to see the graph larger, you can click the magnify icon to see it larger.
- Sysstat: You can check Oracle's stat in minutes. Oracle delivery indicators or EZIS calculated indicators are displayed.
- IO : Displays the IOSTAT_FUNCTION of Oracle. There are two types of choices for this approach.
- I/O Type
Write_reqs is shown in the form (large_Write_reqs + Small_Write_reqs).
Read_reqs is shown in the form (large_Read_reqs + Small_Read_reqs).
When zooming in an area, a “Write MB, Read MB chart” is added.
- I/O Function
Requests = (small_read_reqs + small_write_reqs + large_read_reqs + large_write_reqs)
- Parallel Execution : Displays sysmetric information divided into metric_name as shown below and converted to per second.
- Active Serial Sessions
- Active Parallel Sessions
When zooming in an area, the following indicators are added:
- PQ QC Session Count
- PQ Slave Session Count
- DDL statements parallelized Per Sec
- DML statements parallelized Per Sec
- Queries parallelized Per Sec
- PX downgraded to serial Per Sec
- PX downgraded 1 to 25% Per Sec
- PX downgraded 25 to 50% Per Sec
- PX downgraded 50 to 75% Per Sec
- PX downgraded 75 to 99% Per Sec
- PX operations not downgraded Per Sec
- Services : Displays sysmetric information divided into metric_name as shown below and converted to per second.
- Background Checkpoints Per Sec
- Background CPU Usage Per Sec
- Background Time Per Sec
- Cloud Watch : Shows Cloudwatch information on RDS or EC2 on AWS. This collects only the Cloudwatch information set by Admin, so if you need more collection, please add the collection data in "Admin > Monitoring > StatName > AWS".
- In some cases, Ec2 and RDS collect the same data, and in others. When we select a monitoring server, we automatically determine EC2 and RDS and graph the corresponding indicators.
- Based on the count of Active sessions, the viewtype set in stat is calculated as sigma or delta and charts Max and Avg per minute.
- The following items are shown by separating per second and transactions using Sysmetric metrics:
- per second : sysmetric data
- Logons Per Sec
- User Transaction Per Sec
- Physical Reads Per Sec
- Redo Writes Per Sec
- Tranactions : : sysmetric data
- Logons Per Txn
- User Transaction Per Sec
- Physical Reads Per Txn
- Redo Writes Per Txn
- Shows the "Average Synchronous Single-Block Read Latency" value of Sysmetric.
- The average time to wait for a synchronous latency of a single block read is milliseconds. Synchronous single-block reads are a very accurate way to assess the performance of the storage subsystem. High latency is typically caused by high I / O request loads. If the CPU load is too high, the delay time can increase.