Using Amazon S3 as a source for Amazon DMS - Amazon Database Migration Service
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Using Amazon S3 as a source for Amazon DMS

You can migrate data from an Amazon S3 bucket using Amazon DMS. To do this, provide access to an Amazon S3 bucket containing one or more data files. In that S3 bucket, include a JSON file that describes the mapping between the data and the database tables of the data in those files.

The source data files must be present in the Amazon S3 bucket before the full load starts. You specify the bucket name using the bucketName parameter.

The source data files can be in the following formats:

For source data files in comma-separated value (.csv) format, name them using the following naming convention. In this convention, schemaName is the source schema and tableName is the name of a table within that schema.

/schemaName/tableName/LOAD001.csv /schemaName/tableName/LOAD002.csv /schemaName/tableName/LOAD003.csv ...

For example, suppose that your data files are in mybucket, at the following Amazon S3 path.

s3://mybucket/hr/employee

At load time, Amazon DMS assumes that the source schema name is hr, and that the source table name is employee.

In addition to bucketName (which is required), you can optionally provide a bucketFolder parameter to specify where Amazon DMS should look for data files in the Amazon S3 bucket. Continuing the previous example, if you set bucketFolder to sourcedata, then Amazon DMS reads the data files at the following path.

s3://mybucket/sourcedata/hr/employee

You can specify the column delimiter, row delimiter, null value indicator, and other parameters using extra connection attributes. For more information, see Endpoint settings for Amazon S3 as a source for Amazon DMS.

You can specify a bucket owner and prevent sniping by using the ExpectedBucketOwner Amazon S3 endpoint setting, as shown following. Then, when you make a request to test a connection or perform a migration, S3 checks the account ID of the bucket owner against the specified parameter.

--s3-settings='{"ExpectedBucketOwner": "AWS_Account_ID"}'

Defining external tables for Amazon S3 as a source for Amazon DMS

In addition to the data files, you must also provide an external table definition. An external table definition is a JSON document that describes how Amazon DMS should interpret the data from Amazon S3. The maximum size of this document is 2 MB. If you create a source endpoint using the Amazon DMS Management Console, you can enter the JSON directly into the table-mapping box. If you use the Amazon Command Line Interface (Amazon CLI) or Amazon DMS API to perform migrations, you can create a JSON file to specify the external table definition.

Suppose that you have a data file that includes the following.

101,Smith,Bob,2014-06-04,New York 102,Smith,Bob,2015-10-08,Los Angeles 103,Smith,Bob,2017-03-13,Dallas 104,Smith,Bob,2017-03-13,Dallas

Following is an example external table definition for this data.

{ "TableCount": "1", "Tables": [ { "TableName": "employee", "TablePath": "hr/employee/", "TableOwner": "hr", "TableColumns": [ { "ColumnName": "Id", "ColumnType": "INT8", "ColumnNullable": "false", "ColumnIsPk": "true" }, { "ColumnName": "LastName", "ColumnType": "STRING", "ColumnLength": "20" }, { "ColumnName": "FirstName", "ColumnType": "STRING", "ColumnLength": "30" }, { "ColumnName": "HireDate", "ColumnType": "DATETIME" }, { "ColumnName": "OfficeLocation", "ColumnType": "STRING", "ColumnLength": "20" } ], "TableColumnsTotal": "5" } ] }

The elements in this JSON document are as follows:

TableCount – the number of source tables. In this example, there is only one table.

Tables – an array consisting of one JSON map per source table. In this example, there is only one map. Each map consists of the following elements:

  • TableName – the name of the source table.

  • TablePath – the path in your Amazon S3 bucket where Amazon DMS can find the full data load file. If a bucketFolder value is specified, its value is prepended to the path.

  • TableOwner – the schema name for this table.

  • TableColumns – an array of one or more maps, each of which describes a column in the source table:

    • ColumnName – the name of a column in the source table.

    • ColumnType – the data type for the column. For valid data types, see Source data types for Amazon S3.

    • ColumnLength – the number of bytes in this column. Maximum column length is limited to2147483647 Bytes (2,047 MegaBytes) since an S3 source doesn't support FULL LOB mode. ColumnLength is valid for the following data types:

      • BYTE

      • STRING

    • ColumnNullable – a Boolean value that is true if this column can contain NULL values (default=false).

    • ColumnIsPk – a Boolean value that is true if this column is part of the primary key (default=false).

    • ColumnDateFormat – the input date format for a column with DATE, TIME, and DATETIME types, and used to parse a data string into a date object. Possible values include:

      - YYYY-MM-dd HH:mm:ss - YYYY-MM-dd HH:mm:ss.F - YYYY/MM/dd HH:mm:ss - YYYY/MM/dd HH:mm:ss.F - MM/dd/YYYY HH:mm:ss - MM/dd/YYYY HH:mm:ss.F - YYYYMMdd HH:mm:ss - YYYYMMdd HH:mm:ss.F
  • TableColumnsTotal – the total number of columns. This number must match the number of elements in the TableColumns array.

If you don't specify otherwise, Amazon DMS assumes that ColumnLength is zero.

Note

In supported versions of Amazon DMS, the S3 source data can also contain an optional operation column as the first column before the TableName column value. This operation column identifies the operation (INSERT) used to migrate the data to an S3 target endpoint during a full load.

If present, the value of this column is the initial character of the INSERT operation keyword (I). If specified, this column generally indicates that the S3 source was created by DMS as an S3 target during a previous migration.

In DMS versions prior to 3.4.2, this column wasn't present in S3 source data created from a previous DMS full load. Adding this column to S3 target data allows the format of all rows written to the S3 target to be consistent whether they are written during a full load or during a CDC load. For more information on the options for formatting S3 target data, see Indicating source DB operations in migrated S3 data.

For a column of the NUMERIC type, specify the precision and scale. Precision is the total number of digits in a number, and scale is the number of digits to the right of the decimal point. You use the ColumnPrecision and ColumnScale elements for this, as shown following.

... { "ColumnName": "HourlyRate", "ColumnType": "NUMERIC", "ColumnPrecision": "5" "ColumnScale": "2" } ...

For a column of the DATETIME type with data that contains fractional seconds, specify the scale. Scale is the number of digits for the fractional seconds, and can range from 0 to 9. You use the ColumnScale element for this, as shown following.

... { "ColumnName": "HireDate", "ColumnType": "DATETIME", "ColumnScale": "3" } ...

If you don't specify otherwise, Amazon DMS assumes ColumnScale is zero and truncates the fractional seconds.

Using CDC with Amazon S3 as a source for Amazon DMS

After Amazon DMS performs a full data load, it can optionally replicate data changes to the target endpoint. To do this, you upload change data capture files (CDC files) to your Amazon S3 bucket. Amazon DMS reads these CDC files when you upload them, and then applies the changes at the target endpoint.

The CDC files are named as follows:

CDC00001.csv CDC00002.csv CDC00003.csv ...
Note

To replicate CDC files in the change data folder successfully upload them in a lexical (sequential) order. For example, upload the file CDC00002.csv before the file CDC00003.csv. Otherwise, CDC00002.csv is skipped and isn't replicated if you load it after CDC00003.csv. But the file CDC00004.csv replicates successfully if loaded after CDC00003.csv.

To indicate where Amazon DMS can find the files, specify the cdcPath parameter. Continuing the previous example, if you set cdcPath to changedata, then Amazon DMS reads the CDC files at the following path.

s3://mybucket/changedata

If you set cdcPath to changedata and bucketFolder to myFolder, then Amazon DMS reads the CDC files at the following path.

s3://mybucket/myFolder/changedata

The records in a CDC file are formatted as follows:

  • Operation – the change operation to be performed: INSERT or I, UPDATE or U, or DELETE or D. These keyword and character values are case-insensitive.

    Note

    In supported Amazon DMS versions, Amazon DMS can identify the operation to perform for each load record in two ways. Amazon DMS can do this from the record's keyword value (for example, INSERT) or from its keyword initial character (for example, I). In prior versions, Amazon DMS recognized the load operation only from the full keyword value.

    In prior versions of Amazon DMS, the full keyword value was written to log the CDC data. Also, prior versions wrote the operation value to any S3 target using only the keyword initial.

    Recognizing both formats allows Amazon DMS to handle the operation regardless of how the operation column is written to create the S3 source data. This approach supports using S3 target data as the source for a later migration. With this approach, you don't need to change the format of any keyword initial value that appears in the operation column of the later S3 source.

  • Table name – the name of the source table.

  • Schema name – the name of the source schema.

  • Data – one or more columns that represent the data to be changed.

Following is an example CDC file for a table named employee.

INSERT,employee,hr,101,Smith,Bob,2014-06-04,New York UPDATE,employee,hr,101,Smith,Bob,2015-10-08,Los Angeles UPDATE,employee,hr,101,Smith,Bob,2017-03-13,Dallas DELETE,employee,hr,101,Smith,Bob,2017-03-13,Dallas

Prerequisites when using Amazon S3 as a source for Amazon DMS

To use Amazon S3 as a source for Amazon DMS, your source S3 bucket must be in the same Amazon Region as the DMS replication instance that migrates your data. In addition, the Amazon account you use for the migration must have read access to the source bucket.

The Amazon Identity and Access Management (IAM) role assigned to the user account used to create the migration task must have the following set of permissions.

{ "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "s3:GetObject" ], "Resource": [ "arn:aws:s3:::mybucket*/*" ] }, { "Effect": "Allow", "Action": [ "s3:ListBucket" ], "Resource": [ "arn:aws:s3:::mybucket*" ] } ] }

The Amazon Identity and Access Management (IAM) role assigned to the user account used to create the migration task must have the following set of permissions if versioning is enabled on the Amazon S3 bucket.

{ "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "s3:GetObject", "S3:GetObjectVersion" ], "Resource": [ "arn:aws:s3:::mybucket*/*" ] }, { "Effect": "Allow", "Action": [ "s3:ListBucket" ], "Resource": [ "arn:aws:s3:::mybucket*" ] } ] }

Limitations when using Amazon S3 as a source for Amazon DMS

The following limitations apply when using Amazon S3 as a source:

  • Don’t enable versioning for S3. If you need S3 versioning, use lifecycle policies to actively delete old versions. Otherwise, you might encounter endpoint test connection failures because of an S3 list-object call timeout. To create a lifecycle policy for an S3 bucket, see Managing your storage lifecycle. To delete a version of an S3 object, see Deleting object versions from a versioning-enabled bucket.

  • A VPC-enabled (gateway VPC) S3 bucket is supported in versions 3.4.7 and higher.

  • MySQL converts the time datatype to string. To see time data type values in MySQL, define the column in the target table as string, and set the task's Target table preparation mode setting to Truncate.

  • Amazon DMS uses the BYTE data type internally for data in both BYTE and BYTES data types.

  • S3 source endpoints do not support the DMS table reload feature.

  • Amazon DMS doesn't support Full LOB mode with Amazon S3 as a Source.

The following limitations apply when using Parquet-format files in Amazon S3 as a source:

  • Dates in MMYYYYDD, or DDMMYYYY are not supported for the S3 Parquet Source date-partitioning feature.

Endpoint settings for Amazon S3 as a source for Amazon DMS

You can use endpoint settings to configure your Amazon S3 source database similar to using extra connection attributes. You specify the settings when you create the source endpoint using the Amazon DMS console, or by using the create-endpoint command in the Amazon CLI, with the --s3-settings '{"EndpointSetting": "value", ...}' JSON syntax.

The following table shows the endpoint settings that you can use with Amazon S3 as a source.

Option Description
BucketFolder

(Optional) A folder name in the S3 bucket. If this attribute is provided, source data files and CDC files are read from the path s3://myBucket/bucketFolder/schemaName/tableName/ and s3://myBucket/bucketFolder/ respectively. If this attribute isn't specified, then the path used is schemaName/tableName/.

'{"BucketFolder": "sourceData"}'

BucketName

The name of the S3 bucket.

'{"BucketName": "myBucket"}'

CdcPath The location of CDC files. This attribute is required if a task captures change data; otherwise, it's optional. If CdcPath is present, then Amazon DMS reads CDC files from this path and replicates the data changes to the target endpoint. For more information, see Using CDC with Amazon S3 as a source for Amazon DMS.

'{"CdcPath": "changeData"}'

CsvDelimiter

The delimiter used to separate columns in the source files. The default is a comma. An example follows.

'{"CsvDelimiter": ","}'

CsvNullValue

A user-defined string that Amazon DMS treats as null when reading from the source. The default is an empty string. If you do not set this parameter, Amazon DMS treats an empty string as a null value. If you set this parameter to a string such as "\N", Amazon DMS treats this string as the null value, and treats empty strings as an empty string value.

CsvRowDelimiter

The delimiter used to separate rows in the source files. The default is a newline (\n).

'{"CsvRowDelimiter": "\n"}'

DataFormat

Set this value to Parquet to read data in Parquet format.

'{"DataFormat": "Parquet"}'

IgnoreHeaderRows

When this value is set to 1, Amazon DMS ignores the first row header in a .csv file. A value of 1 enables the feature, a value of 0 disables the feature.

The default is 0.

'{"IgnoreHeaderRows": 1}'

Rfc4180

When this value is set to true or y, each leading double quotation mark has to be followed by an ending double quotation mark. This formatting complies with RFC 4180. When this value is set to false or n, string literals are copied to the target as is. In this case, a delimiter (row or column) signals the end of the field. Thus, you can't use a delimiter as part of the string, because it signals the end of the value.

The default is true.

Valid values: true, false, y, n

'{"Rfc4180": false}'

Source data types for Amazon S3

Data migration that uses Amazon S3 as a source for Amazon DMS needs to map data from Amazon S3 to Amazon DMS data types. For more information, see Defining external tables for Amazon S3 as a source for Amazon DMS.

For information on how to view the data type that is mapped in the target, see the section for the target endpoint you are using.

For additional information about Amazon DMS data types, see Data types for Amazon Database Migration Service.

The following Amazon DMS data types are used with Amazon S3 as a source:

Using Parquet-format files in Amazon S3 as a source for Amazon DMS

In Amazon DMS version 3.5.3 and later, you can use Parquet-format files in an S3 bucket as a source for both Full-Load or CDC replication.

DMS only supports Parquet format files as a source that DMS generates by migrating data to an S3 target endpoint. File names must be in the supported format, or DMS won't include them in the migration.

For source data files in Parquet format, they must be in the following folder and naming convention.

schema/table1/LOAD00001.parquet schema/table2/LOAD00002.parquet schema/table2/LOAD00003.parquet

For source data files for CDC data in Parquet format, name and store them using the following folder and naming convention.

schema/table/20230405-094615814.parquet schema/table/20230405-094615853.parquet schema/table/20230405-094615922.parquet

To access files in Parquet format, set the following endpoint settings:

  • Set DataFormat to Parquet.

  • Do not set the cdcPath setting. Make sure that you create your Parquet-format files in the specified schema/ table folders.

For more information about settings for S3 endpoints, see S3Settings in the Amazon Database Migration Service API Reference.

Supported datatypes for Parquet-format files

Amazon DMS supports the following source and target data types when migrating data from Parquet-format files. Ensure that your target table has columns of the correct data types before migrating.

Source data type Target data type
BYTE BINARY
DATE DATE32
TIME TIME32
DATETIME TIMESTAMP
INT1 INT8
INT2 INT16
INT4 INT32
INT8 INT64
NUMERIC DECIMAL
REAL4 FLOAT
REAL8 DOUBLE
STRING STRING
UINT1 UINT8
UINT2 UINT16
UINT4 UINT32
UINT8 UINT
WSTRING STRING
BLOB BINARY
NCLOB STRING
CLOB STRING
BOOLEAN BOOL