Looker Performance Optimization - Episode 4

Key Concepts:

• Henry (Likely a Looker performance optimization tool)
• Models (In the context of Looker, likely referring to data models)
• Explorers (In the context of Looker, likely referring to data exploration interfaces)
• Unused Explorers
• Query Count
• Python Virtual Environment
• Linux Environment
• henry analyze models command
• henry vacuum models command

Analysis of Looker Models using Henry

Mike DeAngelo, a developer relations engineer at Google, demonstrates the fourth installment of optimizing Looker performance using a tool called Henry. He begins by connecting to a pre-configured Linux environment and navigating to the Henry directory, as established in the first installment.

Setting up the Environment

To begin, Mike activates a Python virtual environment using the command source bin/activate. This ensures that the necessary dependencies for Henry are available.

Analyzing Models with henry analyze models

The core of this installment focuses on analyzing Looker models. Mike executes the command henry analyze models --section=sandbox. He notes that this process will take several minutes to complete. The "No Data Found" messages that may appear during the execution are safe to ignore.

Interpreting the Output of henry analyze models

After the command finishes, Mike explains the output. The output is structured in columns:

1. Project Name: The name of the Looker project.
2. Model Name: The name of the specific Looker model.
3. Number of Explorers: The total number of explorers associated with the model.
4. Number of Unused Explorers: The number of explorers that are not being actively used.
5. Query Count: The number of queries run against the model.

The example shows that most queries in the system are run against masa_thelook_events, indicating it's a heavily used model.

Identifying Unused Explorers with henry vacuum models

To further refine the analysis, Mike uses the henry vacuum models command. Similar to the previous command, this also takes several minutes to run, and "No Data Found" errors can be ignored.

Interpreting the Output of henry vacuum models

The output of henry vacuum models lists the explorers that are currently unused for each model. This information is deemed "very valuable" because it identifies explorers that can potentially be removed entirely, streamlining the Looker environment and potentially improving performance.

Conclusion and Next Steps

The video concludes by highlighting the significance of identifying and potentially removing unused explorers. The fifth and final installment will focus on analyzing and vacuuming explorers directly. The henry vacuum models command provides actionable insights for cleaning up Looker models by pinpointing unused explorers.