Prosumer provers

The prosumer tier sits between hobby (laptop/NAS, < 100 TB) and enterprise (data center, multi-PB). Typical setup: a home rack or a small server in a colocation closet, 100 TB to 5 PB of storage, ~1 Gbps symmetric.

This is the sweet spot for someone who wants to make storage their actual side income or a small business. Earnings scale linearly with bytes proven.

---

Who this is for

• An operator with 100 TB to 5 PB of usable disk
• Symmetric 1 Gbps (residential fibre or business connection)
• Comfortable with Linux, systemd, ZFS or similar redundancy
• Willing to commit 12+ months of operation

What you get

	Prosumer tier
Min usable disk	100 TB
Min stake	100 PROVA × committedGiB (with USDC-equivalent floor)
Recommended hardware	Bare metal server, ECC RAM, ZFS or RAID-Z2, separate redundant network
Setup time	1–2 days (including hardware burn-in)
Realistic capacity	100 TB – 5 PB
Identity attestation	wallet active 90+ days, optionally SBT or ENS subdomain (not full KYC)

Setup overview

The CLI install is the same as hobby tier:

curl -fsSL https://get.prova.network/prover | sh

Differences from hobby tier:

1. Multiple disks via ZFS pool. provad can map multiple physical disks into one logical pool. Configure in prover.toml under [storage].
2. Dedicated network interface. Recommend a separate NIC for proof traffic so retrievals don't impact proof submissions during peak.
3. Rate-monitor your stake. At 1 PB committed, your stake is in the millions of PROVA. Set up alerts when the USDC-equivalent floor moves close to your minimum (the stake-floor oracle).
4. Multi-region announce. If you have presence in multiple regions, register multiple provers (one per region) under separate wallets. Each will earn proportionally.

Hardware notes

Component	Recommended	Notes
CPU	8+ cores, modern x86 or ARM	Storage is disk-bound, not CPU-bound. Older 8-core Xeons work fine.
RAM	64 GB ECC	ZFS likes RAM. ECC catches silent corruption.
Disk	NVMe metadata + spinning rust for bulk	NVMe for the ZFS metadata special-vdev; HDDs for the bulk.
Network	1 Gbps symmetric, real bandwidth	Asymmetric "1 Gbps down / 50 Mbps up" residential is not enough.
Power	UPS, 60+ minute runtime	Sustained outages cause missed proofs and slashing.

Cost estimate for a 1 PB prosumer setup (mid-2026 prices):

• HDDs: $9,000–14,000 (1 PB usable from raw with ZFS Z2)
• NVMe (metadata + L2ARC): $400
• Server (used Dell R730 or similar): $1,500
• Network: $0 (existing fibre) – $200/mo (business circuit)
• Power: ~$50–80/mo
• One-time: $11,000–16,000
• Monthly: $50–280

Earnings model

Same as hobby tier — you earn USDC per deal proven, plus PROVA emission proportional to bytes proven. The emission curve is published; the Earnings page has worked examples for prosumer scale.

A 1 PB prosumer at 70% utilisation, $2.50/TB-month effective price, 99.5% proof success:

USDC monthly:  1 PB × 1024 TB × 0.70 × $2.50 = $1,792 USDC
PROVA monthly: ~ 1 PB / total network committed × monthly_emission

The PROVA emission share scales with your fraction of total committed bytes on the network. Early provers earn a larger absolute share because the network is small; this is intentional.

Slashing exposure

Prosumer tier has substantially more capital at risk than hobby. A 1 PB prover with the default slashFraction = 10% and slashPerFault = 50 PROVA is hit for ~50 PROVA per individual missed proof, and up to 10% of total locked stake per fault event.

Practical implication: at 1 PB you have ~10M PROVA staked. A serious cluster failure (e.g. ZFS pool degraded for 24 h producing missed proofs) could cost up to 1M PROVA. Set up monitoring before you scale above 100 TB.

The protocol's stake-floor oracle gives you a 7-day grace window to top up stake if PROVA's USD price drops; use this window, don't ignore the alert.

Identity attestation

Above 100 TB committed, the protocol requires lightweight identity attestation (not full KYC). Options, any one suffices:

• An ENS name with verified contenthash and an ENS-resolver TXT record claiming this prover wallet
• An EAS attestation (Ethereum Attestation Service) signed by a published attester
• A 30-day continuous on-chain history showing legitimate activity

This is to discourage Sybil — the same person spinning up 50 prosumer provers from the same hardware to capture a larger emission share.

Operational discipline

• 24/7 monitoring. Prometheus + Grafana setup recommended. The official provad Grafana dashboard is included.
• Alerting. Set up PagerDuty / OpsGenie / opsgenie / your phone for: missed proofs, disk errors, network partition, stake-floor breach.
• Quarterly capacity review. Plan disk growth. Adding capacity requires re-registering with the new committed bytes and topping up stake.
• Annual disk replacement budget. HDDs fail. Replace SMART-warning drives within 7 days.

Next steps

• Hardware for the full per-tier table
• Enterprise if you cross 5 PB
• Earnings for the per-tier earnings math