HP P6000 Performance Advisor Software - Manual (Page 30)

Following is an approximate database growth rate for different polling intervals and number of
performance objects based on the tests conducted in the HP lab environment:
Approx database growth rate Number of objects Polling interval
6 GB/month 200 30 Seconds
40 GB/month 2000 45 Seconds
41 GB /month 4096 90 Seconds
For more information on configuring performance and state data collection polling intervals, see
“Adding performance objects for monitoring” (page 55) .
Statistical data
The statistical data shown in HP P6000 Performance Advisor is based on historical data and
includes the basic and advanced statistics. You can view statistical data for storage systems and
performance objects for a predefined or custom duration. By default, the basic statistical data is
displayed based on data collected in the last hour.
IMPORTANT:
•
When performance data is viewed as an aggregate of both controllers, the statistical values
(Standard Deviation, minimum, maximum, average, and percentile) are based on the same
aggregate. It means that both the controllers data is aggregated and then the statistical values
are derived from those aggregated values.
•
Average of a counter (example, percentage or transfer size) for a duration is derived from all
the values. For example: Average of Read Percent is calculated based on the (sum of Read
IOPS of all the performance samples) * 100/(sum of read and write of all the performance
samples) for a specified duration.
Basic statistics
Basic statistical data includes the following:
•
The minimum value sample collected during the selected duration
•
The maximum value sample collected during the selected duration
•
The average value sample collected during the selected duration
Advanced statistics
Advanced statistical data includes the basic statistical data and also the following:
•
Standard Deviation.
Standard deviation is a statistical measure of variability and considered as the root mean
square deviation of the values from their arithmetic mean. A low standard deviation indicates
that the data points are closer to the mean value. A high standard deviation indicates that the
data points are spread over a large range of values. For example, high standard deviation
for any Throughput (Req/s) counter indicates that the host did not receive uniform I/O requests
for the selected duration.
•
90th, 95th, and 99th percentile value samples for the selected duration.
A percentile is a value on a scale of 100 that indicates the percent of distribution equal to the
percentile value. For each data series, the 90th, 95th, and 99th percentile are computed to
ignore the data points in top 10%, 5% or 1% of the utilization peaks. Then, the value is
rounded off to the nearest data sample, so that infrequent peaks are ignored. Such estimates
can help maximize utilization of your systems to ensure only the required storage performance
capacity.
30 HP P6000 Performance Advisor concepts