Bender - Serverless ETL Framework
Alternatives To Bender
Project NameStarsDownloadsRepos Using ThisPackages Using ThisMost Recent CommitTotal ReleasesLatest ReleaseOpen IssuesLicenseLanguage
Choetl65257a month ago181June 24, 202255mitC#
ETL framework for .NET (Parser / Writer for CSV, Flat, Xml, JSON, Key-Value, Parquet, Yaml, Avro formatted files)
Omniparser56112 months ago21September 19, 2022mitGo
omniparser: a native Golang ETL streaming parser and transform library for CSV, JSON, XML, EDI, text, etc.
Nango Sync294
a month ago37otherTypeScript
Sync external APIs to your DB, fast.
a year ago4November 19, 202121mitGo
StorageTapper is a scalable realtime MySQL change data streaming, logical backup and logical replication service
Elastic2424113 days ago13January 11, 20207otherR
R client for the Elasticsearch HTTP API
Awesome Integration216
13 days agocc0-1.0Markdown
A curated list of awesome system integration software and resources.
a year ago14apache-2.0Java
Bender - Serverless ETL Framework
Marklogic Data Hub1271a day ago16May 09, 202239otherTypeScript
The MarkLogic Data Hub: documentation ==>
a year ago12gpl-2.0Java
Gallia Core73
3 months ago4January 05, 20221otherScala
A schema-aware Scala library for data transformation
Alternatives To Bender
Select To Compare

Alternative Project Comparisons


Bender - Serverless ETL Framework This project provides an extendable Java framework for creating serverless ETL functions on AWS Lambda. Bender handles the complex plumbing and provides the interfaces necessary to build modules for all aspects of the ETL process.

An example of what you can do is enrich your log data by appending GeoIP information. For example:

input: {"ip": ""}
output: {"ip": "", "geo_ip": {"location": {"lat": 37.751, "lon": -97.822}}}

This enables you to create dashboards like these in Kibana: Kibana GeoIP


Bender is easily configurable with either json or yaml. The configuration guide provides documentation for option specifics and sample_configs/ contains real world examples of how Bender is configured with commonly used pipelines.

Initial Support

While developers are able to write their own Input Handlers, Deserializers, Operations, Wrappers, Serializers, Transporters, or Reporters, out of box the Bender contains basic functionality to read, filter, manipulate, and write JSON from Amazon Kinesis Streams or Amazon S3 files. Specially the following is supported:


Handlers interface between Amazon Lambda triggers and data sources provided to your ETL function. Events or lines are able to read from:

  • Kinesis
  • S3
  • S3 via SNS
Pre Deserialization Filters

Modular filter support is not yet included but basic string matching and regex based filters are included as a part of Bender Core.


Included is a generic JSON deserializer to transform strings into GSON objects. This allows processing of loosely defined schemas such as those done in application logging. For schema validation the use of GSON to POJOs is encouraged.


Data sometimes needs to be transformed, fixed, sanitized, or enriched. Operations allow for these types of data manipulation.

  • Geo IP lookup
  • JSON Root node promoter (in the case you have nested data)
  • JSON Array Dropping
  • JSON Array Splitting (turning a single event into multiple)
  • Appending value type information to key names (helps with writing data to ElasticSearch)

Optionally wrap data to provide additional information on where the data originated from and what processed it:

  • Kinesis Wrapper
  • S3 Wrapper
  • Basic Wrapper

Write your transformed and wrapped data back into JSON before loading it elsewhere.


Transporters convert string payloads to serialized wire formats and send batches of data to destinations.

  • Firehose
  • S3 (partitioning support included)
  • Elasticsearch (time based index inserts)
  • Splunk
  • Scalyr
  • Sumo Logic
  • Datadog

Monitor the ETL process at each phase within your function and output those metrics for easy consumption in

  • Cloudwatch Metrics
  • Datadog

Local Development/Testing

When you are developing your configuration (or a new transport), you may want to test the configuration with some local data. This can be done with the CLI tool.

The Bender CLI tool operates almost entirely as if it were in Lambda. It mocks out the appropriate source of the event (Kinesis, S3, etc) and passes the data in to the Handler exactly the way Lambda would.

(Note, this only supports the KinesisHandler at the moment).

Command Line Arguments

$ java -jar bender-cli-1.0.0-SNAPSHOT.jar
2018-03-30 11:06:27,482 ERROR - Missing required option: s
usage: BENDER_CONFIG=file://config.yaml java bender-cli-1.0.0-SNAPSHOT.jar
 -H,--help                        Print this message
 -h,--handler <arg>               Which Event Handler do you want to simulate?
                                  Your options are: KinesisHandler.
                                  Default: KinesisHandler
    --kinesis_stream_name <arg>   What stream name should we mimic? Default: log-stream (Kinesis
                                  Handler Only)
 -s,--source_file <arg>           Reference to the file that you want to process. Usage depends on
                                  the Handler you chose. If you chose KinesisHandler then this is a
                                  local file.

Actual processing of data

You can use the Bender CLI tool to actually process data and push it to your destination. This is not the ended production use of the tool, but for validation of your configuration this can work very well.

Here is an example of processing some data, and outputting a simple local file with the new data.


  type: KinesisHandler
  fail_on_exception: true
  add_shardid_to_partitions: true
- name: Syslog Messages
  source_regex: arn:aws:kinesis:.*:.*:stream/.*
    type: GenericJson
    - field: MESSAGE
      prefix_field: MESSAGE_PREFIX
  - type: TimeOperation
    time_field: $.EPOCH
    time_field_type: SECONDS
  - type: JsonKeyNameOperation
  - type: JsonDropArraysOperation
  type: KinesisWrapper
  type: Json
  type: File
  filename: output.json


{"program":"app","host":"","facility":"local0","priority":"info","message":"Some prefix ... {\"type\":\"image\",\"action\":\"pull\",\"time\":1522366659,\"id\":\"image:id\",\"actor\":{\"id\":\"image:id\",\"attributes\":{\"name\":\"image\"}},\"timenano\":1522366659090067620,\"status\":\"pull\"}","epoch":"1522366659.091506"}


$ BENDER_CONFIG=file://config.yaml java -jar bender-cli-1.0.0-SNAPSHOT.jar --source_file data.json
2018-03-30 11:26:24,278 INFO  - Invoking the Kinesis Handler...
2018-03-30 11:26:24,950 INFO  - Parsing
2018-03-30 11:26:25,011 INFO  - Processed 1000 records
2018-03-30 11:26:25,031 INFO  - Bender Initializing (config: file://config.yaml)
2018-03-30 11:26:25,040 DEBUG - Generating BenderConfig object... this can take a little bit
2018-03-30 11:26:30,447 INFO  - Using source: Syslog Messages[sourceRegex=arn:aws:kinesis:.*:.*:stream/.*, containsStrings=[], regexPatterns=[]], deserializers=[GenericJsonDeserializer]], operations=[TimeOperation, KeyNameOperation, DropArraysOperation]]

Output Here's the output of the file.. just the first line:

$ head -1 output.json
{"partition_key":"1","sequence_number":"744","source_arn":"arn:aws:kinesis:us-east-1:123456789:stream/log-stream","function_name":"cli-main","processing_time":1522434390690,"arrival_time":1522434384946,"processing_delay":67731599,"timestamp":1522366659091,"payload":{"program__str":"app","host__str":"","facility__str":"local0","priority__str":"info","message":{"type__str":"image","action__str":"pull","time__long":1522366659,"id__str":"image:id","actor":{"id__str":"image:id","attributes":{"name__str":"image"}},"timenano__long":1522366659090067620,"status__str":"pull"},"message_prefix__str":"Some prefix ... "  ,"epoch__str":"1522366659.091506"}}

Here is the output in a slightly more readable format:

  "partition_key": "1",
  "sequence_number": "744",
  "source_arn": "arn:aws:kinesis:us-east-1:123456789:stream/log-stream",
  "function_name": "cli-main",
  "processing_t  ime": 1522434390690,
  "arrival_time": 1522434384946,
  "processing_delay": 67731599,
  "timestamp": 1522366659091,
  "payload": {
    "program__str": "app",
    "host__str": "",
    "facility__str": "local0",
    "priority__str": "info",
    "message": {
      "type__str": "image",
      "action__str": "pull",
      "time__long": 1522366659,
      "id__str": "image:id",
      "actor": {
        "id__str": "image:id",
        "attributes": {
          "name__str": "image"
      "timenano__long": 1522366659090067700,
      "status__str": "pull"
    "message_prefix__str": "Some prefix ... ",
    "epoch__str": "1522366659.091506"


The easiest way to deploy your function is to use Apex. A sample project is included under example_project/. The project provides an example of a function that is triggered by Kinesis, drops data matching a regex, and forwards the rest to Firehose.

Note to deploy the example you will need to create an IAM role to allow your lambda function to read from kinesis and write to firehose. Your role will need the following two policies:

arn:aws:iam::aws:policy/service-role/AWSLambdaKinesisExecutionRole arn:aws:iam::aws:policy/AmazonKinesisFirehoseFullAccess

After creating your role edit example_project/project.json with the role ARN. You will also need to create the source Kinesis and destination Firehose streams.

To deploy:

cd example_project/
make deploy
DRY=false make deploy


Features and bug fixes are welcome. Please adhere to the following guidelines:

  • Use Google's Java style guide for your IDE.
  • Be conscientious of dependencies you add to Bender Core.
  • Help maintain unit test code coverage by adding tests for each branch in new code.
Popular Etl Projects
Popular Json Projects
Popular Data Processing Categories
Related Searches

Get A Weekly Email With Trending Projects For These Categories
No Spam. Unsubscribe easily at any time.
Aws Lambda
Aws S3
Etl Framework