Add query_hash & query_parameterized_hash columns (#137)

Co-authored-by: Sidharth Pallerla <[email protected]>
get-select · Nov 10, 2023 · 23e1cb0 · 23e1cb0
1 parent 4b00b5b
commit 23e1cb0
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 1 deletion.
diff --git a/models/query_history_enriched.sql b/models/query_history_enriched.sql
@@ -106,6 +106,10 @@ select
     query_history.query_acceleration_bytes_scanned,
     query_history.query_acceleration_partitions_scanned,
     query_history.query_acceleration_upper_limit_scale_factor,
+    query_history.query_hash,
+    query_history.query_hash_version,
+    query_history.query_parameterized_hash,
+    query_history.query_parameterized_hash_version,
 
     -- Rename some existing columns for clarity
     query_history.total_elapsed_time as total_elapsed_time_ms,

diff --git a/models/query_history_enriched.yml b/models/query_history_enriched.yml
@@ -115,6 +115,16 @@ models:
       - name: query_acceleration_upper_limit_scale_factor
         description: Upper limit scale factor that a query would have benefited from.
 
+      # Snowflake's Query Hash & Parameterized Query Hash columns
+      - name: query_hash
+        description: The hash value computed based on the canonicalized SQL text.
+      - name: query_hash_version
+        description: The version of the logic used to compute QUERY_HASH.
+      - name: query_parameterized_hash
+        description: The hash value computed based on the parameterized query.
+      - name: query_parameterized_hash_version
+        description: The version of the logic used to compute QUERY_PARAMETERIZED_HASH.
+
       # Renamed existing columns
       - name: total_elapsed_time_ms
         description: Elapsed time (in milliseconds).

diff --git a/models/staging/stg_query_history.sql b/models/staging/stg_query_history.sql
@@ -63,7 +63,11 @@ select
     is_client_generated_statement,
     query_acceleration_bytes_scanned,
     query_acceleration_partitions_scanned,
-    query_acceleration_upper_limit_scale_factor
+    query_acceleration_upper_limit_scale_factor,
+    query_hash,
+    query_hash_version,
+    query_parameterized_hash,
+    query_parameterized_hash_version
 from {{ source('snowflake_account_usage', 'query_history') }}
 
 {% if is_incremental() %}

diff --git a/models/staging/stg_query_history.yml b/models/staging/stg_query_history.yml
@@ -130,3 +130,11 @@ models:
         description: Number of partitions scanned by the query acceleration service.
       - name: query_acceleration_upper_limit_scale_factor
         description: Upper limit scale factor that a query would have benefited from.
+      - name: query_hash
+        description: The hash value computed based on the canonicalized SQL text.
+      - name: query_hash_version
+        description: The version of the logic used to compute QUERY_HASH.
+      - name: query_parameterized_hash
+        description: The hash value computed based on the parameterized query.
+      - name: query_parameterized_hash_version
+        description: The version of the logic used to compute QUERY_PARAMETERIZED_HASH.