Calibrating an Oracle Cluster Health Advisor Model for a Cluster Deployment

3.4 Calibrating an Oracle Cluster Health Advisor Model for a Cluster Deployment

As shipped with default node and database models, Oracle Cluster Health Advisor is designed not to generate false warning notifications.

You can increase the sensitivity and accuracy of the Oracle Cluster Health Advisor models for a specific workload using the chactl calibrate command.

Oracle recommends that a minimum of 6 hours of data be available and that both the cluster and databases use the same time range for calibration.

The chactl calibrate command analyzes a user-specified time interval that includes all workload phases operating normally. This data is collected while Oracle Cluster Health Advisor is monitoring the cluster and all the databases for which you want to calibrate.

To check if sufficient data is available, run the query calibration command.
If 720 or more records are available, then Oracle Cluster Health Advisor successfully performs the calibration. The calibration function may not consider some data records to be normally occurring for the workload profile being used. In this case, filter the data by using the KPISET parameters in both the query calibration command and the calibrate command.
For example:
```
$ chactl query calibration -db oltpacdb -timeranges 
'start=2016-07-26 01:00:00,end=2016-07-26 02:00:00,start=2016-07-26 03:00:00,end=2016-07-26 04:00:00' 
-kpiset 'name=CPUPERCENT min=20 max=40, name=IOTHROUGHPUT min=500 max=9000' -interval 2
```
Start the calibration and store the model under a user-specified name for the specified date and time range.
For example:
```
$ chactl calibrate cluster –model weekday –timeranges ‘start=2016-07-03 20:50:00,end=2016-07-04 15:00:00’
```
After completing the calibration, Oracle Cluster Health Advisor automatically stores the new model in GIMR.
Use the new model to monitor the cluster as follows:
For example:
```
$ chactl monitor cluster –model weekday
```

Example 3-2 Output for the chactl query calibrate command

Database name : oltpacdb
Start time : 2016-07-26 01:03:10
End time : 2016-07-26 01:57:25
Total Samples : 120
Percentage of filtered data : 8.32%
The number of data samples may not be sufficient for calibration.

1) Disk read (ASM) (Mbyte/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
4.96      0.20      8.98      0.06      25.68   

<25       <50       <75       <100      >=100    
97.50%    2.50%     0.00%     0.00%     0.00%    

2) Disk write (ASM) (Mbyte/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
27.73     9.72      31.75     4.16      109.39  

<50       <100      <150      <200      >=200    
73.33%    22.50%    4.17%     0.00%     0.00%    

3) Disk throughput (ASM) (IO/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
2407.50   1500.00   1978.55   700.00    7800.00 

<5000     <10000    <15000    <20000    >=20000  
83.33%    16.67%    0.00%     0.00%     0.00%    

4) CPU utilization (total) (%)

MEAN      MEDIAN    STDDEV    MIN       MAX     
21.99     21.75     1.36      20.00     26.80   

<20       <40       <60       <80       >=80     
0.00%     100.00%   0.00%     0.00%     0.00%    

5) Database time per user call (usec/call)

MEAN      MEDIAN    STDDEV    MIN       MAX     
267.39    264.87    32.05     205.80    484.57  

<10000000  <20000000  <30000000  <40000000  <50000000  <60000000  <70000000  >=70000000
100.00%   0.00%     0.00%     0.00%     0.00%     0.00%     0.00%     0.00%

Database name : oltpacdb
Start time : 2016-07-26 03:00:00
End time : 2016-07-26 03:53:30
Total Samples : 342
Percentage of filtered data : 23.72%
The number of data samples may not be sufficient for calibration.

1) Disk read (ASM) (Mbyte/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
12.18     0.28      16.07     0.05      60.98   

<25       <50       <75       <100      >=100    
64.33%    34.50%    1.17%     0.00%     0.00%    

2) Disk write (ASM) (Mbyte/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
57.57     51.14     34.12     16.10     135.29  

<50       <100      <150      <200      >=200    
49.12%    38.30%    12.57%    0.00%     0.00%    

3) Disk throughput (ASM) (IO/sec)

MEAN      MEDIAN    STDDEV    MIN       MAX     
5048.83   4300.00   1730.17   2700.00   9000.00 

<5000     <10000    <15000    <20000    >=20000  
63.74%    36.26%    0.00%     0.00%     0.00%    

4) CPU utilization (total) (%)

MEAN      MEDIAN    STDDEV    MIN       MAX     
23.10     22.80     1.88      20.00     31.40   

<20       <40       <60       <80       >=80     
0.00%     100.00%   0.00%     0.00%     0.00%    

5) Database time per user call (usec/call)

MEAN      MEDIAN    STDDEV    MIN       MAX     
744.39    256.47    2892.71   211.45    45438.35

<10000000  <20000000  <30000000  <40000000  <50000000  <60000000  <70000000  >=70000000
100.00%   0.00%     0.00%     0.00%     0.00%     0.00%     0.00%     0.00%