Introduction

A batch source that ingests data from dynamodb into hydrator pipelines.

Use case(s)

• For an organization all the spam emails are being dumped to dynamodb table. As a data scientist I want to train my machine learning models in hydrator pipelines based on the data from the dynamodb tables.

User Storie(s)

• User should be able to provide the table name in DynamoDb.
• User should be able to provide the AWS endpoint url for DynamoDb instance.
• User should be able to provide the AWS region id for DynamoDb instance.
• User should be able to provide the throughput for DynamoDb instance. (Dynamo db charges are incurred based on throughput and user should be able to control the throughput)
• User should be able to provide the AWS access id.
• User should be able to provide the AWS access key.

Plugin Type

•  Batch Source
•  Batch Sink 
•  Real-time Source
•  Real-time Sink
•  Action
•  Post-Run Action
•  Aggregate
•  Join
•  Spark Model
•  Spark Compute

Configurables

This section defines properties that are configurable for this plugin. 

Design / Implementation Tips

• For Testing purposes tables can be created either using AWS cli or using java code http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/JavaDocumentAPITablesExample.html
• AWS dynamoDb cli refrence http://docs.aws.amazon.com/cli/latest/reference/dynamodb/
• Java example for CRUD operations http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/batch-operation-document-api-java.html
• Java Example for working with queries http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/QueryingJavaDocumentAPI.html
• Please reuse/modify RecordReader, InputFormat classes present here https://github.com/awslabs/emr-dynamodb-connector

Design

Dynamo Db JSON Format:

{

  "name": "DynamoDb",

  "type": "batchsource",

    "properties": {

                  "accessKey": "xyz",

                  "secretAccessKey": "abc",

                  "regionId": "us-east-1",

                  "endpointUrl": "localhost:8000",                  "tableName": "Movies",

                  "query": "Id = :v_Id",

                  "filterQuery": "rating > :v_rating",

                  "valueMap": ":v_Id|120,:v_rating|18",

                  "placeholderType": ":v_Id|int,:v_rating|int"

    }

}

Approach(s)

1. Dropdown with the list of regions will be provided to user, to select the region for AWS Dynamo DB to connect to. Supported regions are:         
   "us-gov-west-1", "us-east-1", "us-east-2", "us-west-1", "us-west-2", "eu-west-1", "eu-west-2", "eu-central-1", "ap-south-1","ap-southeast-1", "ap-southeast-2", "ap-northeast-1", "ap-northeast-2", "sa-east-1", "cn-north-1", "ca-central-1", "getCurrentRegion".   (Referred from: http://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/
   http://docs.aws.amazon.com/general/latest/gr/rande.html#ddb_region)
   
   
2.  If user does not select any region, then default region will be used, i.e. us_west_2.
   getCurrentRegion from the list, returns a Region object representing the region the application is running in, when running in EC2. If this method is called from a non-EC2 environment, it will return null.
   
   
3. User will provide the query in the form of keyCondition expression through “Query” widget, that will be used to fetch the items from DynamoDb table. For example: Id = :v_id and rating > :v_rating
   Referred from : http://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/com/amazonaws/services/dynamodbv2/datamodeling/DynamoDBQueryExpression.html#setKeyConditionExpression-java.lang.String-
   
   
4. If there is any filter query required to fetch the data, it can be provided through “Filter Query” widget, in similar fashion the query is provided.
   
5.  User will provide the actualt actual value for the placeholders used in Query/FilterQuery through nameMap and valueMap nameMappings and valueMappings widget respectively.
   
   
6. User will provide the type for the placeholder i.e the attribute values placeholderType widget.
   
   
7. Current implementation supports 'boolean, int, long, double, float and string' types for serachingsearching, i.e attibute value type.

Properties

• endpointUrl: aws endpoint http://docs.aws.amazon.com/general/latest/gr/rande.html#ddb_region This could be reconstructed using regionId.referenceName: This will be used to uniquely identify this source for lineage, annotating metadata, etc.

• accessKey: Access key for AWS DynamoDB to connect to. (Macro Enabled)

• secretAccessKey: Secret access key for AWS DynamoDB to connect to. (Macro Enabled)

• regionId: The region for AWS Dynamo DB DynamoDB to connect to.
• accessKey: Access key for AWS Dynamo DB.
• secretAccessKey: Secret access key for AWS Dynamo DBDefault is us_west_2 i.e. US West (Oregon).

• endpointUrl: The hostname and port for AWS DynamoDB instance to connect to, separated by a colon. For example,localhost:8000.

• tableName: The DynamoDB table to read the data from.
• throughput: Intended throughput for AWS Dynamo DB.

• query: The query that will fetch query: Query to read the items from DynamoDB table. Query must include a partition key value and an equality condition and it must be specified in the following format: 'hashAttributeName = :hashval'. For example, ID = :v_id or ID = :v_id AND Title = :v_title, if sort key condition is used to read the data from table. (Macro Enabled)

• filterQuery: Query that will be used to filter the fetched data, before returning the final results.nameMap: Comma Filter query to return only the desired items that satisfy the condition. All other items are discarded. It must be specified in the similar format like main query. For example, rating = :v_rating. (Macro Enabled)

• nameMappings: List of the placeholder tokens used as the attribute name in the 'Query or FilterQuery' along with their actual attribute names. This is a comma-separated list of key-value pairpairs, where key is the Attribute name place holder used in Query/FilterQuery and value to replace the placeholders.
• valueMap: Comma separated list of key value pair, where key is the value place holder used in Query/FilterQuery for items to be searched and value to replace the placeholders.
• placeholderType:  List of Attribute value placeholders and its type.each pair is separated by a pipe sign '|' and specifies the tokens and actual attribute names. For example, '#yr|year', if the query is like: '#yr = :yyyy'. This might be necessary if an attribute name conflicts with a DynamoDB reserved word. (Macro Enabled)

• valueMappings: List of the placeholder tokens used as the attribute values in the 'Query or FilterQuery' along with their actual values. This is a comma-separated list of key-value pairs, where each pair is separated by a pipe sign '|' and specifies the tokens and actual values. For example, ':v_id|256,:v_title|B', if the query is like: 'ID = :v_id AND Title = :v_title'. (Macro Enabled)

• placeholderType: List of the placeholder tokens used as the attribute values in the 'Query or FilterQuery' along with their data types. This is a comma-separated list of key-value pairs, where each pair is separated by a pipe sign '|' and specifies the tokens and its type. For example, ':v_id|int,:v_title|string', if the query is like: 'ID = :v_id AND Title = :v_title'. Supported types are: 'boolean, int, long, float, double and string'. (Macro Enabled)

• readThroughput: Read Throughput for AWS DynamoDB table to connect to, in double. Default is 1. (Macro Enabled)

• readThroughputPercentage: Read Throughput Percentage for AWS DynamoDB table to connect to. Default is 0.5. (Macro Enabled)

Security

• The AWS access keys should be a password field and macros enabled

Limitation(s)

Future Work

Test Case(s)

DynamoDB batch source - query using partition key.

DynamoDB batch source - query using partition and sort key.

DynamoDB batch source - with query and filter query.

Sample Pipeline

DynamoDBSourcePipeline-cdap-data-pipeline.json

DynamoDBSource_WithSortQuery-cdap-data-pipeline.json

DynamoDBSource_WithFilterQuery-cdap-data-pipeline.json