System Limits

This page documents the hard limits and recommended values for HDDS configuration.

Domain & Participant Limits

Limit	Value	Notes
Domain ID range	0-232	233 domains max (RTPS spec)
Participants per domain	0-119	120 participants max
Total participants	27,960	233 × 120

Stress Test Results

Validated on real hardware:

Test	Participants	Topics	Time	Memory	Status
Scalability	100	1000	11.2s	1.3 GB	✅

Test	Duration	Messages	Rate	Crashes	Leaks	Status
Endurance	12 hours	4,290,256	99.3 msg/s	0	0	✅
Reconnection	20 min	1000 cycles	0.8 cyc/s	0	0	✅
Network partition	30s blackout	701 total	Immediate recovery	0	0	✅
TCP transport	Layer test	5+6 msgs	Bidirectional	0	0	✅

Endurance test memory profile:

• Initial: 4 MB → Peak: 18 MB → Final: 18 MB
• Growth: 14 MB over 12 hours = no memory leak

Memory per participant: ~13 MB (includes discovery, history cache, RTPS state)

Memory Planning

For large deployments, estimate memory as:

Total RAM ≈ Participants × 13 MB + Topics × 0.5 MB

Port Ranges

Port Type	Formula	Domain 0, Participant 0
SPDP Multicast	7400 + (250 × domainId)	7400
SEDP Unicast	7410 + (250 × domainId) + (2 × participantId)	7410
User Unicast	7411 + (250 × domainId) + (2 × participantId)	7411
Data Multicast	7401 + (250 × domainId)	7401

Port range per domain: 7400-7469 (70 ports)

Resource Limits (Default)

Resource	Default	Maximum
max_samples	100,000	Platform memory
max_instances	1	Platform memory
max_samples_per_instance	100,000	≤ max_samples
max_quota_bytes	100 MB	Platform memory

Message Size Limits

Limit	Value	Notes
Max CDR payload	16 MB	Per message
Max UDP datagram	65,507 bytes	Fragmented if larger
RTPS fragment size	64 KB	Default fragmentation

Discovery Timing

Parameter	Default	Adjustable
SPDP announcement period	3 seconds	Yes
SPDP aggressive phase	200 ms × 5	First 5 announcements
Participant lease duration	30 seconds	Yes
Lease check interval	1 second	Internal

History Limits

Configuration	Constraint
KEEP_LAST depth	> 0
KEEP_ALL	Limited by ResourceLimits
Queue validation	max_samples ≥ max_samples_per_instance × max_instances

Identifier Limits

Limit	Value
Topic name length	256 characters
Type name length	256 characters
Partition name	256 characters
UserData size	64 KB

IDL Limits (hdds_gen)

Limit	Value
Identifier length	255 characters
Collection nesting depth	5 levels
Reserved keywords	34 words
Supported annotations	26 standard

Shared Memory Transport

Experimental

Shared memory transport is experimental with significant limitations.

Parameter	Default
Ring buffer slots	256 (power of 2)
Payload size per slot	4 KB
Writer push latency	< 200 ns
Reader poll latency	< 100 ns

SHM Limitations

Limitation	Description
QoS Support	BestEffort only - no ACK/NACK/retransmission
Platforms	Linux/macOS (POSIX shm_open) - Windows not supported
Data Loss	Possible overrun if reader is slow (lock-free ring buffer)
Cross-Vendor	No interop with other DDS vendors (Iceoryx, RTI SHM)
Security	CryptoPlugin bypassed on SHM path

When to Use SHM

✅ Good for:

• High-frequency sensor data (same host)
• BestEffort acceptable scenarios
• Maximum performance (< 1μs latency)
• Single-process / intra-process communication

❌ Not for:

• Reliable delivery requirements
• Cross-host communication
• Mixed vendor environments
• Security-critical data

Message Fragmentation (DATA_FRAG)

Parameter	Value
Default fragment size	1024 bytes
Max unfragmented payload	8192 bytes (8 KB)
Max reassembly size	~64 KB
Fragment buffer timeout	10 seconds

Tested Payload Sizes

Payload	Fragments	Latency	Status
64 B	1 (DATA)	~1.1 ms	Supported
16 KB	16	~1.1 ms	Supported
64 KB	64	~2.3 ms	Supported
128 KB	128	N/A	Exceeds buffer

Large Messages

For payloads > 64 KB, consider increasing the fragment buffer size or using a different transport (TCP, QUIC).

Performance Benchmarks

Metric	Value
Write latency (best)	257 ns
Throughput	4.48 M msg/s
Memory (idle)	< 100 KB
Memory (per topic)	< 50 KB

hdds_viewer Limits

Limit	Value
Frame table buffer	10,000 messages
Timeline buffer	10,000 messages
Undo history	100 levels
Write throughput	625 MB/s
Ingestion rate	41 M msg/s

Recommended Practices

High Frequency Topics (> 1000 Hz)

use hdds::QoS;

let qos = QoS::best_effort().keep_last(1);

Large Messages (> 64 KB)

use hdds::QoS;
use std::time::Duration;

let qos = QoS::reliable()
    .max_blocking_time(Duration::from_secs(5))
    .max_quota_bytes(500 * 1024 * 1024);  // 500 MB

Many Instances (> 1000)

use hdds::QoS;

let qos = QoS::reliable()
    .max_instances(10_000)
    .max_samples_per_instance(10)
    .max_samples(100_000);