Skip to content

Commit

Permalink
doc: update index_recommendations.md
Browse files Browse the repository at this point in the history
The indentation of footnote was incorrect, making it not load. This commit just fix it.

Part Of CRDB-35839

Release note: None
  • Loading branch information
maryliag authored and wenyihu6 committed Feb 21, 2024
1 parent c7e6213 commit 036e083
Showing 1 changed file with 10 additions and 10 deletions.
20 changes: 10 additions & 10 deletions docs/tech-notes/observability/index_recommendations.md
Original file line number Diff line number Diff line change
Expand Up @@ -85,13 +85,13 @@ All options above will display a button to create/alter/drop the index directly
from the Console UI.

[^1]: The cost of generating index recommendations is highly variable. Index recommendations are generated by:
1. Analyzing the query to find "hypothetical indexes" that may improve the performance of the query.
2. Running the query optimizer as if the hypothetical indexes actually exist.
3. Any hypothetical index in the final query plan becomes an index recommendation.

Step 1 is not picky. Its goal is to cover all possible indexes that might help.
In general, the number of hypothetical indexes grows with the number of filtered columns
in the query.
Step 2 can be fast for simple queries (<1ms), but the cost of optimization grows with
the number of joins, filters, and columns (>1s).
If Step 1 adds many hypothetical indexes, Step 2 will take longer because there are more query plans to explore.
1. Analyzing the query to find "hypothetical indexes" that may improve the performance of the query.
2. Running the query optimizer as if the hypothetical indexes actually exist.
3. Any hypothetical index in the final query plan becomes an index recommendation.

Step 1 is not picky. Its goal is to cover all possible indexes that might help.
In general, the number of hypothetical indexes grows with the number of filtered columns
in the query.
Step 2 can be fast for simple queries (<1ms), but the cost of optimization grows with
the number of joins, filters, and columns (>1s).
If Step 1 adds many hypothetical indexes, Step 2 will take longer because there are more query plans to explore.

0 comments on commit 036e083

Please sign in to comment.