Eclipse Ditto 0.9.0

0.9.0

Description

Following the initial release of Ditto, 0.8.0, which was done in 11/2018 the next release 0.9.0 will focus on the following areas:

  • Memory improvements for huge amounts (multi million) of digital twins which are held in memory
  • Adding metrics and logging around the connectivity feature in order to enable being able to operate connections to foreign systems/brokers via APIs
  • Enhancing Ditto's connectivity feature by additionally being able to connect to Apache Kafka
  • Performance improvements of Ditto' search functionality
  • Stabilization of cluster bootstrapping
  • Refactoring of how the services configurations are determined
  • Addition of a Helm template in order to simplify Kubernetes based deployments
  • Contributions from Microsoft in order to ease operating Eclipse Ditto on Microsoft Azure
API Certification

The project leadership certifies that the APIs in this release are "Eclipse Quality".

Architectural Issues

Ditto's codebase is implemented in Java. By default Java objects are mutable which does not work well in highly scalable, message driven architectures, as possible side-effects during runtime may occur when concurrently modifying objects. Because of that the Ditto team decided to use immutable objects wherever and whenever possible. Immutablity of the objects is ensured in unit tests.

Ditto's model (and also the Ditto Protocol) modules are OSGi bundles so that they may be used in OSGi environments (e.g. Eclipse Kura or Eclipse Smarthome) without much effort. The model modules are configured to be checked by a "binary compatibility checker" in the Maven build so that APIs are not broken unintentionally.

On a level higher than the model, Ditto uses a microservice based architecture. As Ditto's microservices are interacting via an event driven approach, Ditto provides a very modular setup on the microservice level meaning that single services must be not started at all if their functionality is not needed. Another benefit from this architecture is that the services may be scaled horizontally if more resources are required.

Ditto utilizes the CQRS and EventSourcing pattern and mainly uses "inserts" (append only) into the database in favor to "updates" in order to get a better performance when doing database writes. The only exception is the "search" microservice which uses traditional CRUD in order to update its search index.

A separate microservice, "connectivity", is responsible for managing and establishing connections to foreign systems. This service is built to be enhanced by new connection types, the currently supported are:

  • AMQP 1.0
  • AMQP 0.9.1 (e.g. RabbitMQ)
  • MQTT 3.1.1
  • Apache Kafka
Security Issues

There are no security issues known at the time of this writing.

Non-Code Aspects

Ditto comes with:

Conforms To UI/UX Guidelines
Not applicable (project doesn't provide UI)
Usability Details

Ditto does not provide a UI yet.

End of Life

Not applicable.

Standards

Ditto is able to process AMQP 1.0, AMQP 0.9.1 and MQTT 3.1.1 messages which are all standardized (IoT) protocols.

Ditto can use JSON Web Tokens specified by RFC 7519 in order to extract "subjects" (e.g. a user-id) to use for the access control to twins.

Communities

The Ditto team is active on meetups, conferences (giving talks), blogs and social media in order to gain visibility in the community.

Users of Ditto ask their questions via these channels:

Ditto seeks for a good integration with Eclipse Hono in order to being able to create digital twins for all devices connected via Hono. The Ditto team is also in regular discussion with the Hono team.

It is difficult to tell how much Ditto is used, one metric we have is the Docker image downloads on Docker hub which are currently ~2.400.