Introduction

RTI Connext supports data types whose size exceeds the maximum message size of the underlying transports. A dds::pub::DataWriter will fragment data samples when required. Fragments are automatically reassembled at the receiving end.

Once all fragments of a sample have been received, the new sample is passed to the dds::sub::DataReader which can then make it available to the user. Note that the new sample is treated as a regular sample at that point and its availability depends on standard QoS settings such as dds::core::policy::ResourceLimits::max_samples and dds::core::policy::HistoryKind::KEEP_LAST.

The large data feature is fully supported by all DDS API's, so its use is mostly transparent. Some additional considerations apply as explained below.

Writing Large Data

In order to use the large data feature with the dds::core::policy::ReliabilityKind::RELIABLE setting, the dds::pub::DataWriter must be configured as an asynchronous writer (rti::core::policy::PublishModeKind_def::ASYNCHRONOUS) with associated rti::pub::FlowController.

While the use of an asynchronous writer and flow controller is optional when using the dds::core::policy::ReliabilityKind::BEST_EFFORT setting, most large data use cases will benefit from the use of a flow controller to prevent flooding the network when fragments are being sent.

Set up writer

Receiving Large Data

Large data is supported by default and in most cases, no further changes are required.

The rti::core::policy::DataReaderResourceLimits allows tuning the resources available to the dds::sub::DataReader for reassembling fragmented large data.

Set up reader