arangodb · ansoboleva · Aug 3, 2022 · Jul 14, 2022 · Jul 15, 2022 · Jul 29, 2022
diff --git a/3.10/aql/invocation-with-arangosh.md b/3.10/aql/invocation-with-arangosh.md
@@ -242,6 +242,44 @@ In the ArangoDB Enterprise Edition there is an additional parameter:
   into sync. The default value is 60.0 (seconds). When the max time has been reached
   the query will be stopped.
 
+Additional parameters for spilling data from the query onto disk
+-----------------------------------------------------------------
+
+  There are two additional parameters that would allow for spilling 
+  intermediate data from the query onto disk for descreasing the memory usage.
+  Note: The option of spilling data from RAM onto disk is experimental and off 
+  by default and this parameter currently only has effect for sorting, meaning 
+  for a query that would use the keyword SORT, but without LIMIT.
+  Also, the query results will still be built up entirely in RAM on coordinators
+  and single servers for non-streaming queries. In order to avoid the buildup of
+  the entire query result in RAM, a streaming query should be used.
+
+
+- `spillOverThresholdNumRows`: This parameter allows for input data and 
+  intermediate results to be spilled onto disk for the execution of a query 
+  after the number of rows reaches the value this parameter holds. This is 
+  used for decreasing the memory usage during the query execution. In a query 
+  that iterates over a collection that contains documents, each row would be a 
+  document and, in a query that iterates over temporary values 
+  (i.e. `FOR i IN 1..100`), each row would be one of such temporary values. 
+  This parameter is experimental and is only taken into account if a path for a
+  directory where the temporary data would be stored is provided with the server 
+  startup option 
+  `--temp.intermediate-results-path`. 
+  Default value: 5000000 rows.
+
+
+- `spillOverThresholdNumRows`: This parameter allows for input data and 
+  intermediate results to be spilled onto disk for the execution of a query 
+  after the memory usage reaches the value in bytes this parameter holds. This 
+  is used for decreasing the memory usage during the query execution. This 
+  parameter is experimental and is only taken into account if a path for a 
+  directory where the temporary data would be stored is provided with the 
+  server startup option 
+  `--temp.intermediate-results-path`. 
+  Default value: 128MB.
+
+
 With _createStatement (ArangoStatement)
 ---------------------------------------
 

diff --git a/3.10/programs-arangod-query.md b/3.10/programs-arangod-query.md
@@ -314,3 +314,63 @@ query results cache, as queries on system collections are internal to ArangoDB a
 only use space in the query results cache unnecessarily.
 
 The default value is *false*.
+
+
+## AQL Query with spilling input data to disk 
+
+With the parameters mentioned below, queries can execute with storing input 
+and intermediate results temporarily on disk for decreasing of memory usage 
+when it reaches a certain threshold. 
+Note: this feature is experimental and off by default. The query results will 
+still be built up entirely in RAM on coordinators and single servers for 
+non-streaming queries. In order to avoid the buildup of the entire query 
+result in RAM, a streaming query should be used.  
+
+The threshold value mentioned to start spilling data onto disk is either 
+the number of rows in the query input or the amount of memory usage in bytes, 
+which are set as query options.
+The main parameter that must be provided for this feature to be active is 
+
+`--temp.intermediate-results-path`. 
+
+This parameter specifies a path to a directory that will be used for temporary
+storage of data. If such path is not provided, the feature of spilling data 
+onto the disk will not be activated.
+Hence, the following parameters would not have effect unless the parameter 
+mentioned above is provided with a directory path.
+The directory specified here must not be located underneath the instance's 
+database directory.
+
+
+`--temp.-intermediate-results-encryption-hardware-acceleration`
+
+Use Intel intrinsics-based encryption, requiring a CPU with the AES-NI 
+instruction set. If turned off, then OpenSSL is used, which may use hardware-
+accelarated encryption too. 
+Default: *true*.
+
+`--temp.intermediate-results-capacity`
+
+Maximum capacity, in bytes to use for ephemeral, intermediate results, meaning 
+the maximum size allowed for the mentioned temporary storage. 
+Default: 0 (unlimited)
+
+`--temp.intermediate-results-encryption` 
+
+Encrypt ephemeral, intermediate results on disk.
+Default: *false*.
+
+
+`--temp.intermediate-results-spillover-threshold-num-rows`: 
+Number of result rows from which on a spillover from RAM to disk will happen.
+Default: *5000000*.
+
+`--temp.intermediate-results-spillover-threshold-memory-usage`: 
+memory usage, in bytes, after which a spillover from RAM to disk will happen.
+Default: *128MB*.
+
+
+
+
+
+
diff --git a/3.10/release-notes-new-features310.md b/3.10/release-notes-new-features310.md
@@ -427,6 +427,53 @@ arangoexport \
   ...
 ```
 
+
+Query changes for decreasing memory usage
+-----------------------------------------
+
+With the parameters mentioned below, queries can execute with storing input
+and intermediate results temporarily on disk for decreasing of memory usage
+when it reaches a certain threshold.
+Note: this feature is experimental and off by default. The query results will
+still be built up entirely in RAM on coordinators and single servers for
+non-streaming queries. In order to avoid the buildup of the entire query
+result in RAM, a streaming query should be used. Also, this feature will
+only take effect currently for AQL SORT operation without LIMIT.
+
+The threshold value mentioned to start spilling data onto disk is either
+the number of rows in the query input or the amount of memory usage in bytes,
+which are set as query options `spillOverThresholdNumRows` and 
+`spillOverThresholdNumRows`.
+
+The main parameter that must be provided for this feature to be active is
+
+`--temp.intermediate-results-path`: specifies a path to a directory that will 
+be used for temporary storage of data. If such path is not provided, the 
+feature of spilling data onto the disk will not be activated.
+Hence, the following parameters would not have effect unless the parameter
+mentioned above is provided with a directory path.
+The directory specified here must not be located underneath the instance's
+database directory.
+
+`--temp.-intermediate-results-encryption-hardware-acceleration`: use Intel 
+intrinsics-based encryption, requiring a CPU with the AES-NI instruction set. 
+If turned off, then OpenSSL is used, which may use hardware-accelarated 
+encryption too. Default: true.
+
+`--temp.intermediate-results-capacity`: maximum capacity, in bytes to use for ephemeral, 
+intermediate results. Default: 0 (unlimited)
+
+`--temp.intermediate-results-encryption`: encrypt ephemeral, intermediate 
+results on disk. Default: false.
+
+`--temp.intermediate-results-spillover-threshold-num-rows`: number of
+result rows from which on a spillover from RAM to disk will happen.
+
+`--temp.intermediate-results-spillover-threshold-memory-usage`: memory
+usage (in bytes) after which a spillover from RAM to disk will happen.

+
+
 Internal changes
 ----------------