System Requirements

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/mystenlabs/sui/llms.txt

Use this file to discover all available pages before exploring further.

​

Validator Requirements

​

Hardware

​

CPU Requirements

• Architecture: x86_64 (AMD64) or ARM64
• Features: AVX2 instruction set recommended for cryptographic operations
• Cores: 24 physical cores minimum (48 virtual cores with hyperthreading)
• Performance: High single-thread performance for consensus operations

​

Memory Requirements

• Minimum: 128 GB RAM
• Recommended: 256 GB RAM for production validators
• Swap: Not recommended; validators should not swap to disk

​

Storage Requirements

• Type: NVMe SSD required (SATA SSDs are too slow)
• Capacity: 4 TB minimum, 8 TB+ recommended
• IOPS: 10,000+ random read/write IOPS
• Throughput: 500+ MB/s sequential read/write
• Latency: Less than 1ms average latency

​

Storage Growth

• Rate: Approximately 50-100 GB per month (network dependent)
• Pruning: Configure authority-store-pruning-config to manage growth
• Planning: Provision at least 12 months of growth capacity

​

Network Requirements

• Bandwidth: 1 Gbps minimum, 10 Gbps recommended
• Latency: Less than 100ms to other validators (lower is better)
• Reliability: Stable connection with less than 1% packet loss
• IPv4: Required (IPv6 optional)

​

Required Ports

​

Fullnode Requirements

​

Hardware

​

CPU Requirements

• Architecture: x86_64 (AMD64) or ARM64
• Cores: 8 cores minimum, 16+ cores recommended
• Features: AVX2 support recommended

​

Memory Requirements

• Minimum: 32 GB RAM
• Recommended: 64 GB RAM for production fullnodes serving RPC traffic
• Swap: Limited swap acceptable (not recommended for performance)

​

Storage Requirements

• Type: SSD required (NVMe recommended for high-traffic nodes)
• Capacity: 2 TB minimum, 4 TB+ recommended
• IOPS: 5,000+ random read/write IOPS
• Throughput: 200+ MB/s sequential read/write

​

Network Requirements

• Bandwidth: 500 Mbps minimum, 1 Gbps recommended
• Latency: less than 200ms to validators (lower is better)
• IPv4: Required

​

Required Ports

​

Operating System

​

Supported Operating Systems

• Linux: Ubuntu 20.04 LTS, Ubuntu 22.04 LTS, Debian 11+ (recommended)
• macOS: macOS 12+ (development only, not recommended for production)
• Windows: Not officially supported

​

Linux Distribution Recommendations

• Ubuntu 22.04 LTS: Recommended for most users
• Debian 11+: Good alternative with long-term support
• RHEL 8+/Rocky Linux: Suitable for enterprise environments

​

Required Software

• Kernel: Linux kernel 5.4+
• libc: glibc 2.31+ or musl libc
• systemd: For service management (recommended)
• Docker: 20.10+ (if using containerized deployment)

​

Environment Variables

# Logging configuration
RUST_LOG=info,sui_core=debug,consensus=debug,jsonrpsee=error
RUST_LOG_JSON=1  # Enable JSON structured logging

# Backtrace on errors
RUST_BACKTRACE=1

# Database configuration
SUI_DB_SYNC_TO_DISK=true  # Enable fsync for durability

# Cache sizing (optional)
SUI_MAX_CACHE_SIZE=100000
SUI_PACKAGE_CACHE_SIZE=1000
SUI_OBJECT_CACHE_SIZE=100000

​

Performance Tuning

​

File Descriptors

# Add to /etc/security/limits.conf
sui soft nofile 65535
sui hard nofile 65535

​

TCP/IP Stack

# Add to /etc/sysctl.conf
net.core.rmem_max = 134217728
net.core.wmem_max = 134217728
net.ipv4.tcp_rmem = 4096 87380 67108864
net.ipv4.tcp_wmem = 4096 65536 67108864
net.ipv4.tcp_congestion_control = bbr

sudo sysctl -p

​

Disk I/O Scheduler

echo none | sudo tee /sys/block/nvme0n1/queue/scheduler

​

Monitoring Requirements

• Prometheus: 20 GB disk, 4 GB RAM
• Grafana: 10 GB disk, 2 GB RAM
• Log aggregation: Size based on retention requirements

​

Backup and Recovery

• Snapshot storage: 2x node database size
• Backup bandwidth: Sufficient for daily snapshots
• Recovery time objective (RTO): less than 4 hours recommended

​

Cloud Provider Recommendations

​

AWS Instance Types

• r6i.8xlarge (32 vCPUs, 256 GB RAM)
• r6i.12xlarge (48 vCPUs, 384 GB RAM)
• Storage: io2 volumes with provisioned IOPS

• m6i.4xlarge (16 vCPUs, 64 GB RAM)
• Storage: gp3 volumes with 3,000+ IOPS

​

Google Cloud Instance Types

• n2-highmem-32 (32 vCPUs, 256 GB RAM)
• n2-highmem-48 (48 vCPUs, 384 GB RAM)
• Storage: SSD persistent disks

• n2-highmem-16 (16 vCPUs, 128 GB RAM)
• Storage: Balanced persistent disks

​

Azure Instance Types

• Standard_E32s_v5 (32 vCPUs, 256 GB RAM)
• Storage: Premium SSD with P50 or higher

• Standard_E16s_v5 (16 vCPUs, 128 GB RAM)
• Storage: Premium SSD

​

Cost Estimation

• Cloud compute: $2,000-4,000/month
• Storage: $500-1,000/month
• Network: $100-500/month
• Total: $2,600-5,500/month

• Cloud compute: $500-1,000/month
• Storage: $200-400/month
• Network: $50-200/month
• Total: $750-1,600/month