Blockchain nodes
Sync and serve Ethereum, Bitcoin, Cosmos or Polkadot with blockchain node hosting built on ECC memory, local NVMe and an unmetered port, paid in BTC, ETH, USDT or LTC.
Execution clients hammer their state database with random reads; local NVMe keeps sync and block processing on pace where HDDs fall behind.
Full nodes upload blocks to peers around the clock, and an unmetered port means serving the network never shows up on an invoice.
Floods against P2P and RPC ports are detected automatically and dropped upstream by always-on L3/L4 mitigation, included with every server.
Run the node here and keep validator keys on a separate signer, linked by a private VLAN if both machines sit in the same city.
Recommended servers
From a first deployment to a busy production setup. Live stock and prices; in-stock picks are online in minutes.
Bitcoin, Cosmos or Polkadot
12C / 24T · 3.7–5.4 GHz
12 Zen 4 cores up to 5.4 GHz and 128 GB DDR5 ECC: a Bitcoin Core node with an Electrum server, or a Cosmos SDK or Polkadot validator.
Our pickEthereum full node or RPC
24C / 48T · 2.85–4.0 GHz
2 × 3.84 TB NVMe keeps an execution and consensus client pair mirrored with room to grow; 256 GB ECC and 10 Gbps serve RPC to your apps.
Solana, archive or multi-chain
32C / 64T · 3.25–3.8 GHz
32 Zen 4 cores, 384 GB DDR5 ECC and 10 Gbps: the class of machine Solana typically needs, or one box for several chains at once.
Sizing
Chain sizes and client requirements change with every release and only grow, so treat these figures as approximate and confirm them against each client’s current documentation.
| Node type | Approximate requirements | Suggested plan |
|---|---|---|
| Bitcoin Core full node + Electrum server | Well over 700 GB of blocks and growing, plus indexes | AMD EPYC 4464P |
| Cosmos SDK or Polkadot validator | Fast cores, pruned state, plus a sentry or backup node | AMD EPYC 4464P |
| Ethereum full node (execution + consensus) | About 2 TB+ of NVMe and growing; 16–32 GB RAM | AMD EPYC 7443P |
| Ethereum RPC node for your own apps | Full-node disk plus RAM and cores for heavy queries | AMD EPYC 7443P |
| Ethereum archive node | Several TB of NVMe, depending heavily on the client | AMD EPYC 9354P |
| Solana validator | Hundreds of GB of ECC RAM, several NVMe, 10 Gbps class | AMD EPYC 9354P |
Expert notes
Short, practical advice from our engineers for this workload.
Chain data only grows, so size for next year rather than for today:
txindex=1 or run an Electrum server. A wallet-only node can set prune=550 and keep only about 550 MB of recent blocks, plus the UTXO set.The EPYC 4464P mirrors 2 × 1.92 TB, which is tight for Ethereum. Take the EPYC 7443P for a mirrored 3.84 TB, or choose RAID 0 at install if you can resync from scratch. Check each client’s current documentation, starting from ethereum.org’s nodes and clients guide: the numbers move with every release.
Your mnemonic and withdrawal credentials should never touch a server. Only signing keys go online, and ideally not on the node itself:
pex = false and persistent_peers pointing at the sentries.Two machines signing with one key is how validators get slashed, and no failover script is worth that risk.
Initial sync is where hardware shows. Consensus clients can start from a trusted checkpoint (--checkpoint-sync-url in Lighthouse) and reach the head quickly; execution clients then sync state from peers, which takes hours to days depending on client and disk. For Bitcoin Core, raise dbcache during the initial block download and lower it afterwards.
127.0.0.1; publish it only through a proxy with authentication and rate limits, or over WireGuard.Root access means you install exactly what you need. These are proven choices for this workload.
FAQ
Plan on roughly 2 TB or more of NVMe for a full node, counting the execution and consensus clients together, and expect growth. Archive nodes need several times that, depending on the client, so check the current documentation for your client pair. The EPYC 7443P gives 3.84 TB mirrored; the EPYC 4464P only fits a full node if you choose RAID 0 for 3.84 TB.
Solana is among the most demanding chains: validators and RPC nodes typically need hundreds of gigabytes of ECC RAM, several fast NVMe drives and a 10 Gbps-class port. The in-stock machine in that class is the EPYC 9354P: 32 cores, 384 GB of DDR5 ECC and a 10 Gbps port. RPC nodes with full account indexes may need more memory, so ask us for a custom build.
Yes, crypto is the only payment method: Bitcoin on-chain and Lightning, Ethereum, USDT on Ethereum or Tron, and Litecoin. Invoices are priced in USD or EUR and the rate is locked for 30 minutes. Delivery starts once the payment confirms, which is near-instant on Lightning and about a minute for USDT on Tron. Renewals can be paid automatically from a prepaid balance.
The network availability target is 99.9% per calendar month per server, with service credits below that, and a failed component is replaced within four hours once diagnosed. For a validator, treat that as a floor, not a plan: keep a synced backup node ready, but never start it with the same keys until the original is confirmed stopped.
Yes, for volumetric attacks: SYN, UDP and amplification floods are filtered at the network level, around the clock and at no extra cost. Abuse that looks like valid requests, such as floods of expensive eth_call or eth_getLogs queries on a public RPC endpoint, is application-layer: rate-limit it in a reverse proxy and keep RPC private where you can.
Pick one of the recommended in-stock servers and it can be online in minutes. Need something bigger or different? Our engineers quote custom builds within one business day.