Patterns and horizontal discovery
A pattern is a series of operations that tell Discovery which CIs to find on your network, what credentials to use, and what tables to populate in the CMDB.
How Discovery uses patterns
A pattern performs the same function as a probe: it identifies and explores a target CI. Discovery uses patterns only during the last two phases of discovery: Identification and Exploration.
When you kick off horizontal discovery with patterns, the Scanning and Classification phases run as they would if you were not using patterns (only using probes and sensors). After the Classification stage completes, Discovery looks at the trigger probes on the classifier to see which probe to launch. When the Horizontal Pattern probe is specified as a trigger probe, Discovery launches both the Horizontal Pattern probe and the pattern that it specifies.
Differences between probes and patterns
The main differences between using patterns for horizontal discovery and using only probes for horizontal discovery are outlined in this table:
| Topic | Horizontal discovery with probes | Horizontal discovery with patterns |
|---|---|---|
| Discovery phases | Probes run during all four discovery phases. | Patterns run only during the identification and exploration phases. Discovery still uses the Shazzam probe and port probes to kick off discovery. |
| Finding new CIs | Discovery offers several out-of-box probes and sensors. However, if you want to find new CIs or change the data that Discovery populates in the CMDB, you must create news probes and sensors or customize the existing ones. To do this, you need knowledge of Java and JavaScript, and you need to understand how probes work together with other probes and with sensors. |
Discovery is offering more patterns in every major release and on the ServiceNow App Store. If you want to create a new pattern or customize an existing one, you can use the pattern designer, which is a user-friendly interface that helps you construct steps and enter values. |
| Feature support | Supports all standard network, and CI discovery. Certain applications are not supported without the use of patterns. See Detailed information on products discovered by ITOM Visibility for a full list. | In addition to supporting all standard discovery, patterns support:
|
| The ECC queue and troubleshooting | Discovery creates Multiple ECC Queue records depending on how many probes are launched. To troubleshoot any issues with horizontal discovery, you must analyze several ECC queue records to see the data that Discovery retrieved. |
Discovery creates only one additional ECC queue record for the pattern. To troubleshoot any issues with horizontal discovery during identification and exploration, you only have to analyze one ECC queue record for the pattern. |
| Performance | Depending on what type of target CI you are trying to identify and explore, Discovery might trigger multiple probes during the identification phase, and then trigger additional probes during the discovery phase. It is possible that horizontal discovery with probes can take longer because of the extra overhead in triggering and processing many probes. | Discovery triggers one pattern during the identification an exploration phases. All operations that Discovery needs to run are in the pattern and Discovery can execute them in order until the pattern is finished. Therefore, pattern discovery can lead to faster performance. |
Patterns for top-down and for horizontal discovery
Both Discovery and Service Mapping can use the same pattern for horizontal and top-down discovery. But they are edited differently. See Create or customize patterns for all steps. If you take a pattern that was exclusively used for top-down discovery and you want to use it for horizontal discovery, you have to make a few modifications. See Use a pattern for horizontal discovery for instructions.
Pattern Orchestrator
Pattern Orchestrator is a process that allows a discovery to trigger multiple patterns related to a collection of data in a hierarchical manner. This is to help address issues with large datasets (or large payloads) that can cause slow discoveries and OOM (out-of-memory) issues on a MID Server.
The Pattern Orchestrator is accessed in the Pattern Orchestrator tab in the Pattern Designer. The tab contains three new tables: Trigger Rules, Orchestrator Input, Orchestrator Output.
- Count must specify a number for the Batch Size Count. The child pattern is triggered each time the parent pattern sends the specified number of CIs.
- All uses every CI sent from the parent pattern and triggers the child one time.
- Use Parent Context Only relies on the parent pattern's output settings to dictate which child pattern should trigger. This is useful when two patterns have the same parent, but you need one to trigger before the other.
The Orchestrator Input uses the CI Type field to accept CI types from its parent pattern or parent's parent pattern. The Saved As field saves the CI type's name to be used in the pattern. Service accounts are a special case and must be named service_account. The CI Attributes field uses any outputs from the parent pattern to select which attributes to use.
The Orchestrator Output controls which data will be sent to subsequent patterns. It has the fields CI Class Type, which is the type of CI being discovered, and Variable Names which specifies the variables to be passed on as inputs to child patterns. Any variables not specified in Variable Names are not sent to child patterns. The variable source_native_key can be manually added by searching for it in the Variable Names field.
Source Native Key
In some situations, one pattern may get partial information about a CI, while another pattern gets partial info about the same CI. Instead of creating redundant records or discarding partial information, the information is stored in the CMDB IRE Partial Payloads [cmdb_ire_partial_payloads_list] table. After enough patterns gather partial information, it can correlate that data to provide complete information on the CI.
Pagination
Another way to reduce stress on the MID Server's memory is pagination. Pagination breaks a large payload up into smaller, ordered slices called pages. Pagination limits the maximum size of items received and generates a token. Once that page is processed, the token is passed to the next page, which is then processed, until all the pages are completed.
Pagination is controlled from the Custom Operations table [sa_custom_operation]. The field Is pagination capable is hidden by default on the form, and can be added to the form with the Personalize Form controls. Once that is set to true, the associated pattern in the pattern designer has the field Paginated. Setting the Paginated field to true runs that step in the pattern with pagination.
Add the Horizontal Pattern probe to a classifier
To use a pattern for the identification and exploration phases of horizontal discovery, you must add the Horizontal Pattern probe to the classifiers for the CIs are you trying to discover.
始める前に
このタスクについて
手順
-
From the related list view, double click the field under the
Pattern column and add the pattern you want to
associate with the classification.
Use a pattern for horizontal discovery
If you want to use a new pattern, or if you already have a pattern that you were using for top-down discovery, you can use the pattern for horizontal discovery with a few modifications to the relevant classifier.
始める前に
このタスクについて
手順
次のタスク
Run the pattern in Debug mode to test it. When you are sure the pattern works, you can run discovery by setting up a discovery schedule or running an on-demand discovery. See Schedule a horizontal discovery for more information.