Why it matters: Consensus nodes perform intensive cryptographic operations. A slow CPU causes the node to fall behind and trigger consensus failures.
Avoid:
ARM processors (not supported)
Consumer-grade CPUs (inadequate cache, no ECC support)
Low-frequency server CPUs (single-thread performance matters)
Memory Requirements
Attribute
Specification
Minimum Capacity
256 GB
Recommended
320 GB+
Type
DDR4 ECC Registered (3200MHz) or DDR5 ECC
ECC
MANDATORY
Why ECC is mandatory: A single bit flip in the in-memory state causes the node to compute a different state hash than other validators, triggering consensus failure. At the scale of 256 GB with constant activity, soft errors are statistically inevitable without ECC.
What happens with insufficient memory:
Heap Size
Observed Behavior
6 GB
Out of memory within hours
16 GB
Out of memory after ~5 days
48 GB
Stable at low TPS with tuning
192-256 GB
Required for mainnet loads
Storage Requirements
Boot Volume (OS):
2x 240 GB SSD minimum in RAID 1 (mirror)
Purpose: Operating system, logs, management tools
Data Volume (Blockchain State):
Attribute
Specification
Capacity
5 TB minimum, 8+ TB recommended
Type
NVMe SSD (enterprise-grade)
Configuration
RAID 0 or RAID 10
Sequential Write
2,000+ MB/s
Random Read IOPS
250,000+
Storage Growth Projections:
Network Load
Daily Growth
Year 1 Total
10 TPS (testnet)
~500 MB
~200 GB
50 TPS (low mainnet)
~3 GB
~1.1 TB
200 TPS (moderate)
~15 GB
~5.5 TB
Network Requirements
Attribute
Specification
Bandwidth
1 Gbps sustained (not burstable)
IP Address
Static IPv4, publicly routable
NAT
Not supported
Latency
<100ms to other validators
Required Ports:
Port
Protocol
Direction
Purpose
50111
TCP
In + Out
Gossip protocol (node-to-node)
50211
TCP
Ingress
gRPC API
50212
TCP
Ingress
TLS-encrypted gRPC
123
UDP
In + Out
NTP time synchronization
Hosting Requirements
VPS vs. Bare Metal: Why VPS is NOT Acceptable
Factor
VPS Risk
Consensus Impact
CPU steal
Noisy neighbors
GC pauses extend -> node falls behind -> failure
Memory overcommit
Hypervisor swapping
Catastrophic performance degradation
Disk I/O contention
Shared storage
State writes stall -> consensus failure
Network jitter
Virtual networking
Gossip delays -> consensus divergence
Acceptable Hosting Options:
Option
Verdict
Notes
Dedicated bare metal (Hetzner, OVH, Equinix)
Recommended
Full hardware isolation
Cloud bare metal (AWS i3en.metal, Azure LSv3)
Acceptable
No hypervisor, but expensive
Colocation
Ideal for large operators
Own hardware, rented rack space
Standard VPS / cloud VM
NOT ACCEPTABLE
Causes failures
Data Center Requirements:
Tier III or higher (N+1 redundancy)
Redundant power (UPS + generator)
Redundant cooling
Physical security (badge access, cameras)
Operating System
Attribute
Specification
Supported OS
Ubuntu 22.04 LTS (64-bit)
Alternative
RHEL 8/9, Oracle Linux 8/9
Kernel
5.15.x or 6.x LTS
Required Software:
Component
Version
Purpose
Docker Engine
20.10.6+
Container runtime
Docker Compose
1.29.2+
Container orchestration
JQ
1.5+
JSON processing
cURL
7.58+
API health checks
Step 1: Contact Onyx DAO and Apply
What you do
Reach out to Onyx DAO on Telegram.
Share your organization name and validator interest.
Provide a technical and operational point of contact.
What Onyx DAO does
Opens your onboarding thread.
Shares the onboarding checklist and target timeline.
Schedules kickoff.
Exit criteria
Your onboarding request is accepted and tracked.
Step 2: Kickoff and Planning
What you do
Join the kickoff telegram group.
Confirm your team roles (technical owner, operations owner, on-call owner).
Confirm preferred deployment windows (UTC).
What Onyx DAO does
Explains the full validator deployment path.
Defines responsibilities between your team and Onyx DAO.
Confirms communication and incident escalation channels.