Each server can produce hundreds of kilobytes of performance data per minute, though the amount of data captured has little to no impact on the performance of the monitored system. By default, the PingDataMetrics server stores performance data for 20 years. Configure the volume of performance data collected by each monitored server so that the PingDataMetrics server can keep up with the flow.

The performance data model is a dimensional data model. Measurements can be taken on multiple simultaneous values that are distinguished by dimension values. For example, a response time metric provides the time in milliseconds it took a server to respond to an LDAP request. This response-time metric has two dimensions:

Application name
Reflects the connection criteria of the request.
Operation type
Corresponds to the LDAP operation, such as add, bind, or search. If a server has 20 different connection criteria, each response-time sample may have 140 different values, one for each of the applications multiplied by the number of operation types.

The performance data captured on the monitored server has a record with the following fields.

Name Data type Description

Timestamp

Date

Time of measurement, using clock on the monitored server

Metric

String

Name of metric

Dimension

String

Values of dimensions 1 - 3

Count

Int

Number of measurements represented by this sample

Average

Double

Average value of this sample

Minimum

Double

Optional minimum value of this sample

Maximum

Double

Optional maximum value of this sample

Buckets

Int

Optional histogram data associated with this sample

When a performance record is imported into the PingDataMetrics server, it is normalized to reduce the size of the record. The normalized record contains the following information.

Name Data type Description

batchID

Int

The ID of the batch of data to which this record belongs

sampleTime

Timestamp

The time the sample was captured or equivalent information after aggregation

metric_qual

Int

The ID of a structure that reflects the metric and all dimension values

definitionID

Int

ID of the histogram definition, if the data belong to a histogram-valued sample

count

Int

Number of measurements represented by this sample

avg_val

Real

Average value for this sample

min_val

Real

Minimum value for this sample

max_val

Real

Maximum value for this sample

val 1-15

Long

Histogram bucket values