S3Settings
Settings for exporting data to Amazon S3.
Contents
- BucketFolder
-
An optional parameter to set a folder name in the S3 bucket. If provided, tables are created in the path
bucketFolder/schema_name/table_name/
. If this parameter isn't specified, then the path used isschema_name/table_name/
.Type: String
Required: No
- BucketName
-
The name of the S3 bucket.
Type: String
Required: No
- CdcInsertsAndUpdates
-
A value that enables a change data capture (CDC) load to write INSERT and UPDATE operations to .csv or .parquet (columnar storage) output files. The default setting is
false
, but whenCdcInsertsAndUpdates
is set totrue
ory
, only INSERTs and UPDATEs from the source database are migrated to the .csv or .parquet file.For .csv file format only, how these INSERTs and UPDATEs are recorded depends on the value of the
IncludeOpForFullLoad
parameter. IfIncludeOpForFullLoad
is set totrue
, the first field of every CDC record is set to eitherI
orU
to indicate INSERT and UPDATE operations at the source. But ifIncludeOpForFullLoad
is set tofalse
, CDC records are written without an indication of INSERT or UPDATE operations at the source. For more information about how these settings work together, see Indicating Source DB Operations in Migrated S3 Data in the AWS Database Migration Service User Guide..Note AWS DMS supports the use of the
CdcInsertsAndUpdates
parameter in versions 3.3.1 and later.CdcInsertsOnly
andCdcInsertsAndUpdates
can't both be set totrue
for the same endpoint. Set eitherCdcInsertsOnly
orCdcInsertsAndUpdates
totrue
for the same endpoint, but not both.Type: Boolean
Required: No
- CdcInsertsOnly
-
A value that enables a change data capture (CDC) load to write only INSERT operations to .csv or columnar storage (.parquet) output files. By default (the
false
setting), the first field in a .csv or .parquet record contains the letter I (INSERT), U (UPDATE), or D (DELETE). These values indicate whether the row was inserted, updated, or deleted at the source database for a CDC load to the target.If
CdcInsertsOnly
is set totrue
ory
, only INSERTs from the source database are migrated to the .csv or .parquet file. For .csv format only, how these INSERTs are recorded depends on the value ofIncludeOpForFullLoad
. IfIncludeOpForFullLoad
is set totrue
, the first field of every CDC record is set to I to indicate the INSERT operation at the source. IfIncludeOpForFullLoad
is set tofalse
, every CDC record is written without a first field to indicate the INSERT operation at the source. For more information about how these settings work together, see Indicating Source DB Operations in Migrated S3 Data in the AWS Database Migration Service User Guide..Note AWS DMS supports the interaction described preceding between the
CdcInsertsOnly
andIncludeOpForFullLoad
parameters in versions 3.1.4 and later.CdcInsertsOnly
andCdcInsertsAndUpdates
can't both be set totrue
for the same endpoint. Set eitherCdcInsertsOnly
orCdcInsertsAndUpdates
totrue
for the same endpoint, but not both.Type: Boolean
Required: No
- CdcPath
-
Specifies the folder path of CDC files. For an S3 source, this setting is required if a task captures change data; otherwise, it's optional. If
CdcPath
is set, AWS DMS reads CDC files from this path and replicates the data changes to the target endpoint. For an S3 target if you setPreserveTransactions
totrue
, AWS DMS verifies that you have set this parameter to a folder path on your S3 target where AWS DMS can save the transaction order for the CDC load. AWS DMS creates this CDC folder path in either your S3 target working directory or the S3 target location specified byBucketFolder
andBucketName
.For example, if you specify
CdcPath
asMyChangedData
, and you specifyBucketName
asMyTargetBucket
but do not specifyBucketFolder
, AWS DMS creates the CDC folder path following:MyTargetBucket/MyChangedData
.If you specify the same
CdcPath
, and you specifyBucketName
asMyTargetBucket
andBucketFolder
asMyTargetData
, AWS DMS creates the CDC folder path following:MyTargetBucket/MyTargetData/MyChangedData
.For more information on CDC including transaction order on an S3 target, see Capturing data changes (CDC) including transaction order on the S3 target.
Note This setting is supported in AWS DMS versions 3.4.2 and later.
Type: String
Required: No
- CompressionType
-
An optional parameter to use GZIP to compress the target files. Set to GZIP to compress the target files. Either set this parameter to NONE (the default) or don't use it to leave the files uncompressed. This parameter applies to both .csv and .parquet file formats.
Type: String
Valid Values:
none | gzip
Required: No
- CsvDelimiter
-
The delimiter used to separate columns in the .csv file for both source and target. The default is a comma.
Type: String
Required: No
- CsvNoSupValue
-
This setting only applies if your Amazon S3 output files during a change data capture (CDC) load are written in .csv format. If
UseCsvNoSupValue
is set to true, specify a string value that you want AWS DMS to use for all columns not included in the supplemental log. If you do not specify a string value, AWS DMS uses the null value for these columns regardless of theUseCsvNoSupValue
setting.Note This setting is supported in AWS DMS versions 3.4.1 and later.
Type: String
Required: No
- CsvRowDelimiter
-
The delimiter used to separate rows in the .csv file for both source and target. The default is a carriage return (
\n
).Type: String
Required: No
- DataFormat
-
The format of the data that you want to use for output. You can choose one of the following:
-
csv
: This is a row-based file format with comma-separated values (.csv). -
parquet
: Apache Parquet (.parquet) is a columnar storage file format that features efficient compression and provides faster query response.
Type: String
Valid Values:
csv | parquet
Required: No
-
- DataPageSize
-
The size of one data page in bytes. This parameter defaults to 1024 * 1024 bytes (1 MiB). This number is used for .parquet file format only.
Type: Integer
Required: No
- DatePartitionDelimiter
-
Specifies a date separating delimiter to use during folder partitioning. The default value is
SLASH
. Use this parameter whenDatePartitionedEnabled
is set totrue
.Type: String
Valid Values:
SLASH | UNDERSCORE | DASH | NONE
Required: No
- DatePartitionEnabled
-
When set to
true
, this parameter partitions S3 bucket folders based on transaction commit dates. The default value isfalse
. For more information about date-based folder partitoning, see Using date-based folder partitioning.Type: Boolean
Required: No
- DatePartitionSequence
-
Identifies the sequence of the date format to use during folder partitioning. The default value is
YYYYMMDD
. Use this parameter whenDatePartitionedEnabled
is set totrue
.Type: String
Valid Values:
YYYYMMDD | YYYYMMDDHH | YYYYMM | MMYYYYDD | DDMMYYYY
Required: No
- DictPageSizeLimit
-
The maximum size of an encoded dictionary page of a column. If the dictionary page exceeds this, this column is stored using an encoding type of
PLAIN
. This parameter defaults to 1024 * 1024 bytes (1 MiB), the maximum size of a dictionary page before it reverts toPLAIN
encoding. This size is used for .parquet file format only.Type: Integer
Required: No
- EnableStatistics
-
A value that enables statistics for Parquet pages and row groups. Choose
true
to enable statistics,false
to disable. Statistics includeNULL
,DISTINCT
,MAX
, andMIN
values. This parameter defaults totrue
. This value is used for .parquet file format only.Type: Boolean
Required: No
- EncodingType
-
The type of encoding you are using:
-
RLE_DICTIONARY
uses a combination of bit-packing and run-length encoding to store repeated values more efficiently. This is the default. -
PLAIN
doesn't use encoding at all. Values are stored as they are. -
PLAIN_DICTIONARY
builds a dictionary of the values encountered in a given column. The dictionary is stored in a dictionary page for each column chunk.
Type: String
Valid Values:
plain | plain-dictionary | rle-dictionary
Required: No
-
- EncryptionMode
-
The type of server-side encryption that you want to use for your data. This encryption type is part of the endpoint settings or the extra connections attributes for Amazon S3. You can choose either
SSE_S3
(the default) orSSE_KMS
.Note For the
ModifyEndpoint
operation, you can change the existing value of theEncryptionMode
parameter fromSSE_KMS
toSSE_S3
. But you can’t change the existing value fromSSE_S3
toSSE_KMS
.To use
SSE_S3
, you need an AWS Identity and Access Management (IAM) role with permission to allow"arn:aws:s3:::dms-*"
to use the following actions:-
s3:CreateBucket
-
s3:ListBucket
-
s3:DeleteBucket
-
s3:GetBucketLocation
-
s3:GetObject
-
s3:PutObject
-
s3:DeleteObject
-
s3:GetObjectVersion
-
s3:GetBucketPolicy
-
s3:PutBucketPolicy
-
s3:DeleteBucketPolicy
Type: String
Valid Values:
sse-s3 | sse-kms
Required: No
-
- ExternalTableDefinition
-
Specifies how tables are defined in the S3 source files only.
Type: String
Required: No
- IncludeOpForFullLoad
-
A value that enables a full load to write INSERT operations to the comma-separated value (.csv) output files only to indicate how the rows were added to the source database.
Note AWS DMS supports the
IncludeOpForFullLoad
parameter in versions 3.1.4 and later.For full load, records can only be inserted. By default (the
false
setting), no information is recorded in these output files for a full load to indicate that the rows were inserted at the source database. IfIncludeOpForFullLoad
is set totrue
ory
, the INSERT is recorded as an I annotation in the first field of the .csv file. This allows the format of your target records from a full load to be consistent with the target records from a CDC load.Note This setting works together with the
CdcInsertsOnly
and theCdcInsertsAndUpdates
parameters for output to .csv files only. For more information about how these settings work together, see Indicating Source DB Operations in Migrated S3 Data in the AWS Database Migration Service User Guide..Type: Boolean
Required: No
- ParquetTimestampInMillisecond
-
A value that specifies the precision of any
TIMESTAMP
column values that are written to an Amazon S3 object file in .parquet format.Note AWS DMS supports the
ParquetTimestampInMillisecond
parameter in versions 3.1.4 and later.When
ParquetTimestampInMillisecond
is set totrue
ory
, AWS DMS writes allTIMESTAMP
columns in a .parquet formatted file with millisecond precision. Otherwise, DMS writes them with microsecond precision.Currently, Amazon Athena and AWS Glue can handle only millisecond precision for
TIMESTAMP
values. Set this parameter totrue
for S3 endpoint object files that are .parquet formatted only if you plan to query or process the data with Athena or AWS Glue.Note AWS DMS writes any
TIMESTAMP
column values written to an S3 file in .csv format with microsecond precision.Setting
ParquetTimestampInMillisecond
has no effect on the string format of the timestamp column value that is inserted by setting theTimestampColumnName
parameter.Type: Boolean
Required: No
- ParquetVersion
-
The version of the Apache Parquet format that you want to use:
parquet_1_0
(the default) orparquet_2_0
.Type: String
Valid Values:
parquet-1-0 | parquet-2-0
Required: No
- PreserveTransactions
-
If set to
true
, AWS DMS saves the transaction order for a change data capture (CDC) load on the Amazon S3 target specified byCdcPath
. For more information, see Capturing data changes (CDC) including transaction order on the S3 target.Note This setting is supported in AWS DMS versions 3.4.2 and later.
Type: Boolean
Required: No
- RowGroupLength
-
The number of rows in a row group. A smaller row group size provides faster reads. But as the number of row groups grows, the slower writes become. This parameter defaults to 10,000 rows. This number is used for .parquet file format only.
If you choose a value larger than the maximum,
RowGroupLength
is set to the max row group length in bytes (64 * 1024 * 1024).Type: Integer
Required: No
- ServerSideEncryptionKmsKeyId
-
If you are using
SSE_KMS
for theEncryptionMode
, provide the AWS KMS key ID. The key that you use needs an attached policy that enables AWS Identity and Access Management (IAM) user permissions and allows use of the key.Here is a CLI example:
aws dms create-endpoint --endpoint-identifier value --endpoint-type target --engine-name s3 --s3-settings ServiceAccessRoleArn=value,BucketFolder=value,BucketName=value,EncryptionMode=SSE_KMS,ServerSideEncryptionKmsKeyId=value
Type: String
Required: No
- ServiceAccessRoleArn
-
The Amazon Resource Name (ARN) used by the service access IAM role. It is a required parameter that enables DMS to write and read objects from an S3 bucket.
Type: String
Required: No
- TimestampColumnName
-
A value that when nonblank causes AWS DMS to add a column with timestamp information to the endpoint data for an Amazon S3 target.
Note AWS DMS supports the
TimestampColumnName
parameter in versions 3.1.4 and later.DMS includes an additional
STRING
column in the .csv or .parquet object files of your migrated data when you setTimestampColumnName
to a nonblank value.For a full load, each row of this timestamp column contains a timestamp for when the data was transferred from the source to the target by DMS.
For a change data capture (CDC) load, each row of the timestamp column contains the timestamp for the commit of that row in the source database.
The string format for this timestamp column value is
yyyy-MM-dd HH:mm:ss.SSSSSS
. By default, the precision of this value is in microseconds. For a CDC load, the rounding of the precision depends on the commit timestamp supported by DMS for the source database.When the
AddColumnName
parameter is set totrue
, DMS also includes a name for the timestamp column that you set withTimestampColumnName
.Type: String
Required: No
- UseCsvNoSupValue
-
This setting applies if the S3 output files during a change data capture (CDC) load are written in .csv format. If set to
true
for columns not included in the supplemental log, AWS DMS uses the value specified byCsvNoSupValue
. If not set or set tofalse
, AWS DMS uses the null value for these columns.Note This setting is supported in AWS DMS versions 3.4.1 and later.
Type: Boolean
Required: No
See Also
For more information about using this API in one of the language-specific AWS SDKs, see the following: