DataPower insight pack supports analysis of default log and extended latency message log produced by the IBM DataPower appliance. The objective of the analysis is to enable administrators to understand the basic health of the system and find transaction bottlenecks, if any.
The insight pack includes two SourceTypes DataPower_DefaultLog and DataPower_ExtLatency along with dashboards for the two types.
Insight Pack Installation and Configuration
The insight pack can be installed (like any other pack) using the pkg_mgmt utility shipped with the Log Analysis server. After installation, an administrator needs to create data sources for the two types and update the custom app files to point to his data sources. Similarly, the time filter (which defaults to 1 day) should be changed to time window required for the analysis.
The mechanism to collect logs from the appliance has to be worked out by the DataPower administrator. From the Log Analysis point of view, one can choose from different loading and streaming data options such as batch loading using the Java data collector or streaming using Log File Agent shipped with Log Analysis or other mechanism such as Logstash.
Analysis of Default log
Default logs are analyzed to understand the basic health of the system by doing the following:
- Analyze objects which are producing the log messages
- Get distribution of records by Log Category
- Get all records for a transaction
- Understand the Message Level trends (info/debug/error).
The insight pack generates the following dashboard to support this analysis. No special configuration is required on the DataPower appliance for the default logs. The insight pack works with the out-of-box log produced.
Analysis of Extended Latency Messaging log
These logs are analyzed to find bottlenecks and issues related to transactions running on the DataPower appliance. An administrator can use the insight pack to do the following:
- Stitch together a transaction executing across multiple services.
- Estimate total time taken by a transaction.
- Show trend of total time taken by different transactions.
- Find distribution of time taken by each of the services in a given transaction.
- Find hot spots (slowest service) in a transaction.
- Generate alert when total time for a transaction crosses a threshold.
The following dashboard is produced by the insight pack to support the above use cases.
To use this source type, the DataPower administrator needs to configure logging for the extended latency message logs for Target Type File, Log Format csv. This csv file produced can then be analyzed by the DataPower_ExtLatency source type.
Access the IBM DataPower Insight Pack for more information and to download the pack.