DDS: Data distribution in real-time systems
Part of the Open Architecture Field Guide.
The Data Distribution Service (DDS) is an Object Management Group standard for data-centric publish-subscribe communication. Applications exchange typed data through topics and specify delivery behavior through quality-of-service policies.
In this guide
One position report, several consumers
A navigation application produces position reports. A map display needs those reports to place a symbol, a recorder needs them for later analysis, and another application may use them in a calculation. If the producer maintains a separate application-specific connection to every consumer, each added consumer becomes another communication dependency for the producer to manage.
The Data Distribution Service (DDS) uses a publish–subscribe model. The producer publishes data through the middleware, and consumers declare the data they want to receive. DDS provides the data model, discovery mechanisms, and delivery policies that support this arrangement. The Object Management Group defines the standard. [1]
This changes how the applications describe their communication needs. The navigation application can be written around producing a position report, while the map and recorder are written around consuming it. The middleware handles distribution between the participating applications according to the configured interfaces and policies.
Topics, types, and samples
A DDS topic associates a name with a data type. The type defines the fields and values in an update. In the navigation example, an illustrative report type could contain a position and its measurement time. That definition allows both the display and recorder to interpret each report consistently.
A DataWriter publishes updates for a topic. Each update is called a sample. A DataReader receives the samples to which it subscribes. These are software entities used by the applications; they separate the application's production or use of information from the work of communicating it.
Discovery allows the communication software to find participating writers and readers. The shared data space describes the information accessible through this model. It is a logical communication arrangement, rather than a requirement for one central database or broker through which every sample must pass. [1]
Scroll horizontally to inspect the diagram.
The diagram follows one publication to several consumers. Each consumer uses the same defined information for its own purpose. Adding a consumer therefore creates a new use of the topic without requiring the navigation algorithm to incorporate that consumer's application logic.
Matching delivery behavior to the application
The display and recorder illustrate why data distribution needs more than a topic name. A display may be interested primarily in the latest position. A recorder may need a sequence of earlier reports. DDS quality-of-service (QoS) policies express the delivery behavior associated with these needs.
History governs which samples are retained, while durability concerns the availability of retained data to readers that join later. Reliability describes the delivery behavior used for samples. A deadline expresses the expected interval between updates, giving the application a way to describe how regularly data should be produced or received. [1]
These policies are considered together. Retaining a long history consumes resources; delivering every update has different implications from keeping only the latest state. The engineering decision begins with what each consumer needs to accomplish and translates that need into communication policies.
The resulting contract connects the publisher and subscriber configurations. Integration testing can then examine actual behavior: whether a newly started consumer obtains the expected data, how a delayed update is handled, and whether the application receives reports within its intended timing conditions.
From the data model to network exchanges
DDS defines the application-facing communication model. The DDS Interoperability Wire Protocol, commonly called DDSI-RTPS, specifies messages and behavior used in communication between DDS implementations. It describes how the data-distribution model is realized in exchanges over a network. [2]
The information carried by those exchanges can come from a separate domain model. The Universal Command and Control Interface (UCI), for example, defines mission-message content, while DDS supplies distribution services. The report's meaning and the mechanism that delivers it remain separate engineering concerns.
For the navigation application, that separation makes the complete chain easier to describe. A measurement or state model defines what the position report means. A topic and type identify the data being shared. Writers, readers, and QoS policies define how applications publish and consume it. The wire protocol connects those participants through the communication infrastructure.