QoS Policies Overview

DDS provides 22 Quality of Service policies to fine-tune data distribution behavior. HDDS implements all standard policies from DDS v1.4.

Most Important Policies

Policy	Description	Default
Reliability	Delivery guarantee	BEST_EFFORT
Durability	Data persistence	VOLATILE
History	Sample buffering	KEEP_LAST(10)
Deadline	Update frequency	Infinite
Liveliness	Failure detection	AUTOMATIC

All 22 QoS Policies

Data Distribution

Policy	Purpose
Reliability	BEST_EFFORT vs RELIABLE delivery
History	KEEP_LAST(n) vs KEEP_ALL buffering
Durability	VOLATILE, TRANSIENT_LOCAL, PERSISTENT
DurabilityService	Late-joiner delivery configuration
ResourceLimits	Max samples, instances, memory

Timing

Policy	Purpose
Deadline	Maximum time between samples
LatencyBudget	Acceptable delay hint
Lifespan	Data expiration time
TimeBasedFilter	Minimum separation between samples

Ownership

Policy	Purpose
Ownership	SHARED vs EXCLUSIVE writer control
OwnershipStrength	Priority for exclusive ownership

Ordering

Policy	Purpose
DestinationOrder	BY_RECEPTION_TIMESTAMP vs BY_SOURCE_TIMESTAMP
Presentation	Coherent/ordered access scope

Lifecycle

Policy	Purpose
Liveliness	AUTOMATIC, MANUAL_BY_PARTICIPANT, MANUAL_BY_TOPIC
WriterDataLifecycle	Auto-dispose unregistered instances
ReaderDataLifecycle	Auto-purge samples without writers
EntityFactory	Auto-enable child entities

Metadata

Policy	Purpose
UserData	Custom user metadata
TopicData	Topic-level metadata
GroupData	Publisher/Subscriber metadata
Partition	Logical data separation
TransportPriority	Network QoS hint

Quick Start

use hdds::{Participant, QoS, TransportMode};

let participant = Participant::builder("my_app")
    .domain_id(0)
    .with_transport(TransportMode::UdpMulticast)
    .build()?;

// Create writer with fluent QoS builder
let writer = participant
    .topic::<SensorData>("sensors/temperature")?
    .writer()
    .qos(QoS::reliable().keep_last(10).transient_local())
    .build()?;

QoS Presets

HDDS provides built-in presets for common scenarios:

use hdds::{Participant, QoS, TransportMode};

let participant = Participant::builder("my_app")
    .domain_id(0)
    .with_transport(TransportMode::UdpMulticast)
    .build()?;

// Low latency (gaming, real-time control)
let writer = participant
    .topic::<ControlData>("control/commands")?
    .writer()
    .qos(QoS::best_effort().keep_last(1))
    .build()?;

// High throughput (bulk data transfer)
let writer = participant
    .topic::<BulkData>("data/transfer")?
    .writer()
    .qos(QoS::best_effort().keep_last(1000))
    .build()?;

// Reliable (guaranteed delivery with caching)
let writer = participant
    .topic::<StateData>("state/updates")?
    .writer()
    .qos(QoS::reliable().keep_last(100).transient_local())
    .build()?;

QoS Compatibility

Writers and readers must have compatible QoS to communicate:

Writer	Reader	Result
reliable()	reliable()	Match
reliable()	best_effort()	Match
best_effort()	reliable()	No Match
best_effort()	best_effort()	Match

Compatibility

A BEST_EFFORT writer cannot satisfy a RELIABLE reader. The reader expects acknowledgments that the writer won't send.

Fluent Builder API

HDDS uses a fluent builder pattern for QoS configuration:

use hdds::QoS;

// Start with a preset
let qos = QoS::reliable()
    .keep_last(100)        // History: keep last 100 samples
    .transient_local();    // Durability: cache for late joiners

// Or start from scratch
let qos = QoS::best_effort()
    .keep_last(1)          // Minimal history
    .volatile();           // No caching

Available Methods

Method	Description
QoS::reliable()	Guaranteed delivery with ACK/NACK
QoS::best_effort()	Fire and forget
.keep_last(n)	Keep last N samples per instance
.keep_all()	Keep all samples
.volatile()	No persistence
.transient_local()	In-memory cache for late joiners
.persistent()	Disk persistence

Next Steps

• Reliability - RELIABLE vs BEST_EFFORT
• Durability - Data persistence
• History - Sample buffering