Nakadi is a distributed event bus broker that implements a RESTful API abstraction on top of Kafka-like queues.
More detailed information can be found on our website.
The goal of Nakadi (ნაკადი means stream in Georgian) is to provide an event broker infrastructure to:
-
Abstract event delivery via a secured RESTful API.
This allows microservices teams to maintain service boundaries, and not directly depend on any specific message broker technology. Access can be managed individually for every queue and secured using OAuth and custom authorization plugins.
-
Enable convenient development of event-driven applications and asynchronous microservices.
Event types can be defined with Event type schemas and managed via a registry. All events will be validated against the schema before publishing. This guarantees data quality and consistency for consumers.
-
Efficient low latency event delivery.
Once a publisher sends an event using a simple HTTP POST, consumers can be pushed to via a streaming HTTP connection, allowing near real-time event processing. The consumer connection has keepalive controls and support for managing stream offsets using subscriptions.
Read more to understand The big picture Architecture for data integration
Watch the talk Data Integration in the World of Microservices
Nakadi is high-load production ready. Zalando uses Nakadi as its central Event Bus Service. Nakadi reliably handles the traffic from thousands event types with the throughput of more than hundreds gigabytes per second. The project is in active development. See the CHANGELOG.md
- Stream:
- REST abstraction over Kafka-like queues.
- CRUD for event types.
- Event batch publishing.
- Low-level interface.
- manual client side partition management is needed
- no support of commits
- High-level interface (Subscription API).
- automatic redistribution of partitions between consuming clients
- commits should be issued to move server-side cursors
- Schema:
- Schema registry.
- Several event type categories (Undefined, Business, Data Change).
- Several partitioning strategies (Random, Hash, User defined).
- Event enrichment strategies.
- Schema evolution.
- Events validation using an event type schema.
- Security:
- OAuth2 authentication.
- Per-event type authorization.
- Blacklist of users and applications.
- Operations:
- STUPS platform compatible.
- ZMON monitoring compatible.
- SLO monitoring.
- Timelines.
- This allows transparently switch production and consumption to different cluster (tier, region, AZ) without moving actual data and any service degradation.
- Opens possibility for implementation of other streaming technologies and engines besides Kafka (like AWS Kinesis, Google pub/sub etc.)
Read more about latest development in our CHANGELOG.md
- Support for different streaming technologies and engines. Nakadi currently uses Apache Kafka as its broker, but other providers (such as Kinesis) will be possible.
- Filtering of events for subscribing consumers.
- Store old published events forever using transparent fall back backup shortages like AWS S3.
- Separate the internal schema register to standalone service.
- Use additional schema formats and protocols like Avro, protobuf and others.
The zalando-nakadi organisation contains many useful related projects like
- Client libraries
- SDK
- GUI
- DevOps tools and more
You can run the project locally using Docker.
The Nakadi server is a Java 8 Spring Boot application. It uses Kafka 0.10.2 as its broker and PostgreSQL 9.5 as its supporting database.
Nakadi requires recent versions of docker and docker-compose. In particular, docker-compose >= v1.7.0 is required. See Install Docker Compose for information on installing the most recent docker-compose version.
The project is built with Gradle.
The ./gradlew
wrapper script will bootstrap
the right Gradle version if it's not already installed.
To get the source, clone the git repository.
git clone https://github.com/zalando/nakadi.git
The gradle setup is fairly standard, the main tasks are:
./gradlew build
: run a build and test./gradlew clean
: clean down the build
Some other useful tasks are:
./gradlew acceptanceTest
: run the ATs./gradlew fullAcceptanceTest
: run the ATs in the context of Docker./gradlew startNakadi
: build Nakadi and start docker-compose services: nakadi, postgresql, zookeeper and kafka./gradlew stopNakadi
: shutdown docker-compose services./gradlew startStorages
: start docker-compose services: postgres, zookeeper and kafka (useful for development purposes)./gradlew stopStorages
: shutdown docker-compose services
For working with an IDE, the eclipse
IDE task is available and you'll be able to import the build.gradle
into Intellij IDEA directly.
From the project's home directory you can start Nakadi via Gradle:
./gradlew startNakadi
This will build the project and run docker compose with 4 services:
- Nakadi (8080)
- PostgreSQL (5432)
- Kafka (9092)
- Zookeeper (2181)
To stop the running Nakadi:
./gradlew stopNakadi
Please read the manual for the full API usage details.
The Nakadi API allows the publishing and consuming of events over HTTP. To do this the producer must register an event type with the Nakadi schema registry.
This example shows minimal undefined
category event type with a wilcard schema -
curl -v -XPOST http://localhost:8080/event-types -H "Content-type: application/json" -d '{
"name": "order.ORDER_RECEIVED",
"owning_application": "order-service",
"category": "undefined",
"schema": {
"type": "json_schema",
"schema": "{ \"additionalProperties\": true }"
}
}'
Note: This is not recommended category and schema. It should be used only for the testing.
Read mode in the manual
You can open a stream for an Event Type via the events
sub-resource:
curl -v http://localhost:8080/event-types/order.ORDER_RECEIVED/events
HTTP/1.1 200 OK
{"cursor":{"partition":"0","offset":"82376-000087231"},"events":[{"order_number": "ORDER_001"}]}
{"cursor":{"partition":"0","offset":"82376-000087232"}}
{"cursor":{"partition":"0","offset":"82376-000087232"},"events":[{"order_number": "ORDER_002"}]}
{"cursor":{"partition":"0","offset":"82376-000087233"},"events":[{"order_number": "ORDER_003"}]}
You will see the events when you publish them from another console for example.
The records without events
field are Keep Alive
messages.
Note: This is the low-level API should be used only for debugging. It is not recommended for production systems. For production systems please use Subscriptions API
Events for an event type can be published by posting to its "events" collection:
curl -v -XPOST http://localhost:8080/event-types/order.ORDER_RECEIVED/events \
-H "Content-type: application/json" \
-d '[{
"order_number": "24873243241",
}, {
"order_number": "24873243242",
}]'
HTTP/1.1 200 OK
Read more in the manual
Nakadi accepts contributions from the open-source community.
Please read CONTRIBUTING.md.
Please also note our CODE_OF_CONDUCT.md.
This email address serves as the main contact address for this project.
Bug reports and feature requests are more likely to be addressed if posted as issues here on GitHub.
Please read the full LICENSE
The MIT License (MIT) Copyright © 2015 Zalando SE, https://tech.zalando.com
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.