Crypto node hosting is one of those workloads that quietly punishes cut corners. A full node isn't a web app that sits idle between requests — it syncs a growing chain to disk, gossips with dozens of peers around the clock, and, if you're validating, it cannot miss its attestation window. These are notes from the rack on running a full node or validator 24/7 on a VPS: what actually bottlenecks, how we size per chain, and the point where you should stop fighting a VPS and move to bare metal.
What crypto node hosting actually stresses
Three resources decide whether a node stays healthy. First, disk IOPS during sync — the initial block download and, worse, the ongoing state writes. Chains like Ethereum's execution layer thrash random reads and writes against a multi-hundred-gigabyte state trie; spinning disks or network-attached block storage fall behind and never catch up. Every VPS tier we sell runs on enterprise SSD in ZFS pools, with no per-VM IOPS cap, for exactly this reason. Second, bandwidth — a node maintains many simultaneous peer connections, and both directions matter because you're serving blocks and state to peers as much as pulling them. Metered or asymmetric links throttle you into peer-count limits. Third, uptime and clock discipline for validators, where a few minutes offline is a missed reward or a penalty.
Our VPS lineup is built around the first two: SSD on every tier with no per-VM IOPS cap, and unmetered symmetric traffic (shaped 100–300 Mbps by plan) so peer traffic in and out never counts against a cap. You get root SSH, KVM virtualization, a dedicated IPv4 (plus free IPv6 in Romania, the Netherlands and Germany), and a deploy of about a minute when you need to spin a second node fast.
Why fast SSD and symmetric bandwidth are non-negotiable
Sync time is dominated by disk. On local SSD an execution-layer sync that would take days on slower storage finishes in a fraction of the time, and — more importantly — the node keeps pace with the chain head instead of drifting behind under normal write pressure. For archive workloads the IOPS demand is relentless, which is where a VPS eventually hits its ceiling (more on that below).
Bandwidth is the other half. A well-connected node holds dozens of peers; each peer relationship is bidirectional gossip. Symmetric unmetered means your upload — the direction most consumer and metered plans starve — is as fat as your download. That keeps your peer count high, your propagation fast, and, for a validator, your attestations landing on time. Full connectivity details for AS60982 are on our network page.
Static IPs and our own AS
Every instance ships with a static dedicated IPv4, plus free IPv6 in Romania, the Netherlands and Germany. That matters more for nodes than for most workloads — peers remember you, some validator setups and monitoring pin to an address, and a static endpoint keeps your node discoverable instead of churning identities on every restart.
Underneath, your address is announced from our own AS60982, which reaches the internet through IP transit from one upstream carrier; there is no IX peering today. If you already announce your own space, our BGP and BYOIP options let you bring your own IPv4 /24 to a dedicated server and have it announced on AS60982, or run a BGP session from your own ASN.
Per-chain sizing
Sizing a node is mostly about disk headroom and RAM for the client's caches. We don't publish per-chain byte counts because they climb every month — check your client's current requirements — but here's how our tiers map to the shape of the work.
| Node profile | What it needs | Where to start |
|---|---|---|
| Light / pruned node (single chain) | Modest RAM, SSD, steady peers | VPS II — 4GB / 2 vCPU / 40GB |
| Full node whose chain data fits the disk | More RAM for state cache, larger disk | VPS III — 8GB / 2 vCPU / 60GB, plus extra storage |
| Validator + full node pair | Headroom so consensus and execution don't contend | VPS IV — 16GB / 4 vCPU / 80GB |
| Heavy full node / multiple chains | Top-tier RAM, CPU, and disk | VPS VI — 32GB / 8 vCPU / 120GB |
Check disk before anything else: the largest VPS carries 120GB of SSD plus up to 150GB of extra storage in 50GB increments. A chain whose full data set is bigger than that belongs on dedicated hardware, whatever the RAM needs look like.
Because billing is hourly, you can bring up a bigger tier for the initial sync, confirm the node is healthy, then copy the synced data to a fresh instance sized for its steady-state footprint and delete the big one (a stopped VPS still bills; deleting it stops the meter), without paying for a month you didn't use. In place, the client panel upgrades a VPS to a bigger shape. The full ladder and pricing sit on the VPS hosting page.
Running a validator 24/7
A validator node in Europe lives or dies by two things a VPS controls well: uninterrupted uptime and low, stable latency to the rest of the network. We back every instance with a 99.9% monthly availability target (service credit on request) and 100 Gbps of always-on Layer 3/4 DDoS protection, so volumetric attacks against your endpoint are mitigated automatically, with nothing for you to switch on mid-duty. Keep your keys on the box you control, keep the clock disciplined with NTP, and let the symmetric link keep your attestations propagating. Static IPs mean your validator's network identity stays put across reboots and client upgrades.
When to jump to bare metal
A VPS is the right tool for most full nodes and validators. Archive nodes are where it stops being the right tool. An archive node keeps every historical state and hammers disk with sustained random IOPS that a shared virtual host can't guarantee — you want dedicated NVMe and no noisy neighbors. That's the signal to move to dedicated hardware: single-tenant machines from a Dual Xeon E5-2630 v3 (16c/32t, 128GB RAM, SSD) up to Dual EPYC 9754 (256c/512t, 1TB RAM, NVMe), with ports from 1 up to 200 Gbps, one IPv4 included (a /31; /29–/27 subnets as add-ons), and BYOIP as free order options: your own /24 announced on AS60982, or your own ASN and /24 over a BGP session. Multiple heavy chains on one operator, or an RPC endpoint serving real traffic, are the same story — the guaranteed resources of dedicated hardware beat fighting for IOPS on a VPS.
How Noded Can Help
We run our own network — AS60982, SSD on every VPS tier, unmetered symmetric traffic, and 100 Gbps DDoS protection included on every product — which is most of what a node operator is shopping for anyway. Start on a VPS sized to your chain, upgrade to a bigger shape from the client panel as the chain grows, and move to dedicated when an archive node or RPC endpoint outgrows shared IOPS. If you already run your own address space, we can announce it on AS60982 from a dedicated server through our BYOIP path — and we're happy to talk sizing before you commit.
FAQ
Can I run an Ethereum node on a VPS?
Not a mainnet full node. RAM is fine from VPS III (8GB / 2 vCPU) upward, but disk is the limit: our largest VPS carries 120GB of SSD plus up to 150GB of extra storage, well short of what a mainnet Ethereum execution and consensus client pair stores today. Run that on dedicated hardware; a VPS suits testnets, light clients and smaller chains whose data fits.
Is a VPS enough for a Bitcoin full node?
For a pruned Bitcoin node, yes — the workload is lighter on IOPS than an Ethereum execution node, it keeps only recent blocks on disk, and a small SSD VPS with our unmetered symmetric bandwidth handles the peer traffic without issue. An unpruned node's full block history is larger than the 120GB plus up to 150GB of extra storage a VPS can carry, so run that one on dedicated hardware.
Do you support running a validator node in Europe?
Yes. Our VPS run in six EU locations (plus Ashburn, US East), with a static IPv4 per instance (free IPv6 in Romania, the Netherlands and Germany), a 99.9% availability target, and 100 Gbps of DDoS protection included — the conditions a validator needs to keep attestations landing on time. Keep your keys on the instance you control.
When should I move a node from a VPS to bare metal?
Move when you're running an archive node, several heavy chains on one machine, or a public RPC endpoint under real load. Those need sustained, guaranteed IOPS and no noisy neighbors — that's dedicated hardware territory, starting at EUR209/mo, with ports up to 200 Gbps and BYOIP or a BGP session with AS60982 as order options.
Do node instances come with static IPs?
Every VPS ships with a static IPv4, plus free IPv6 in Romania, the Netherlands and Germany. That keeps your node discoverable to peers across reboots and client upgrades, and gives validators and monitoring a fixed endpoint to pin to.