Concepts
What is a Package?
A deployment entity which contains definitions (classes, view, objects, dashboard, PDF reports).
A package is a collection of pre-configured elements that can be imported into a ServicePilot configuration. A number of parameters are applied to the package when importing it into the configuration to monitor equipment as desired. These parameters are package dependent but a graphical wizard presents the information requested.
ServicePilot provides a host of build-in packages to monitor different equipment types. If a package does not exist for a particular device, it is often possible to monitor this device using a collection of built-in packages to monitor different parts of the device’s operation.
Example: A server-microsoft-windows-snmp package defines components of a Windows Server to monitor (Disk, CPU, Memory, Ethernet, etc) and how to discover and monitor each of these using SNMP.
What is a Resource?
Use of package.
A resource is one use of a package in the running configuration. The package definition plus the unique parameters applied in the package wizard result in a unique resource that helps build up the elements monitored by ServicePilot.
Example: A server-microsoft-windows-snmp package is added to the ServicePilot configuration and becomes a resource to be monitored by the provisioning of this package with specific package parameters such as SNMP access credentials.
What are Views?
Hierarchical containers to organize resources by location or business function.
Objects containing statistics are added to ServicePilot by placing them in a logical hierarchy of elements called views. Each uniquely named view may contain other views or objects as well as other graphical elements. The nested views allow different hierarchies to be built up by ServicePilot administrators, possibly to present equipment by geographic location or service grouping.
The worst status of the monitored equipment percolates up from the indicators to their container objects and then on to the view in which the object is placed, all the way up the hierarchy to the root MAIN view.
Example: An Office A view might be created as a container in which to place all servers, network switches and applications that are based at this particular location.
What are Objects?
A collection of statistics.
Objects are uniquely named collections of indicators, each indicator being a captured or calculated statistic. Each object will collect data from a single source. Based on the availability of source data, object availability over time is calculated. If any indicators exceed defined thresholds then the object will also take on the state of the worst performing indicator. Objects may therefore be in the following states:
| State | Description | |
|---|---|---|
 |
OK | Object source data is available and all of the object’s indicators are nominal or unknown. |
 |
Minor | At least one of the object’s indicators has passed a minor threshold but none of the indicators are currently major or critical. |
 |
Major | At least one of the object’s indicators has passed a major threshold but none of the indicators are currently critical. |
 |
Critical | At least one of the object’s indicators has passed a critical threshold or the object is in a critical service anomaly state. |
 |
Unavailable | The object source data is unreachable or the object is in an unavailable service anomaly state. |
 |
Unknown | The object’s state has not been determined either because it has yet to collect data or it has specifically been told to ignore data by placing the object in an unmanaged state. An unknown state can also be the result of all indicator thresholds being set to unknown. |
Example: An object named Server - eth0 collects data from a single Ethernet interface.
What are Classes?
Object technical properties (definition of the collection, indicators, thresholds, …).
Classes regroup a series of properties forming the basis of a family of like objects. Each object is based on only one class definition, however multiple objects can be based on the same class. Classes define what indicators an object will have, the type of data expected for each indicator and how this data is calculated before being stored. Classes can also contain scripts to discover elements to be monitored and indicator statistics calculated from captured data. Finally, classes also specify the default indicator thresholds and capture frequency for objects.
Example: An ‘Interface’ class defines what objects based on this class will collect and their status thresholds. Each object based on this class will behave in the same way apart from collecting the information from a different source.
What is an Indicator?
A single statistic.
Each indicator in an object presents a single statistic over time.
Indicators are either collected from monitored equipment or calculated:
| Indicator type | Source |
|---|---|
| Monitor | The indicator data is obtained by querying the monitored device. |
| Complex | The indicator data is calculated based on the values of the monitored indicator data received. |
| Information | An SNMP indicator or table that can be retrieved on demand. |
| Control | An SNMP indicator used to check that values retrieved for the object still correspond to the same monitored element. |
Example: An object named Server - eth0 includes indicators ‘Status’ and ‘In Traffic’. Indicators like the ‘In Load’ percentage are then calculated.
What is a Threshold?
A threshold is a condition giving a status to an indicator.
Each indicator can have up to 3 fixed thresholds, 1 dynamic threshold and 2 Machine Learning thresholds that will change the status of the indicator, the object and its view hierarchy.
| Thresholds type | Details |
|---|---|
| Fixed thresholds | These are three successive operators, three corresponding states and three fixed values that are tested against the indicator value. If no value matches, the state of the Else field is applied to the indicator. |
| Spike threshold | If the indicator value deviates abruptly from its expected value, a spike threshold exception will be raised. The Spike status is calculated only if a critical or unavailable fixed threshold is present on the indicator and a regression line can be determined. |
| Daily trend threshold | Every 15 minutes daily indicator trends are calculated for the last 24 hours of data. If the indicator’s daily trend is matched by the daily trend Operator and Value then a daily trend threshold alert is raised. |
| Monthly trend threshold | Every day monthly indicator trends are calculated for the last 30 days of data. If the indicator’s monthly trend is matched by the monthly trend Operator and Value then a monthly trend threshold alert is raised. |
| Critical alert prediction time threshold | If the predicted time when an indicator is to exceed the critical threshold is less that the specified alert Value then a prediction alert will be raised. |
| Dynamic threshold | Based on a script, determines whether or not it is necessary to activate the indicator thresholds. |
What are Events?
Unsolicited data receipt.
ServicePilot may be sent SNMP Traps, Notifications, Syslogs or other custom log data. On receipt of events, additional meta-data can be added before storage in data collections by event type. This resource can then be queried and filtered as well as performing statistical analysis to produce graphs, lists and alerts.
Example: A SNMP Trap is received by ServicePilot. It is decoded and tagged with a ‘Severity’ label and stored in the Trap database collection.
What are Policies?
Common configuration applied to resources. Specified in resources or views.
See the Policies documentation for details.
