Three ways to own Solana compute
Twelve server shapes, three isolation models, one decision: how much of the silicon do you need to own? Shared vCPU from $100, pinned cores with a dedicated private VLAN to the NLN fleet from $245, the whole machine from $458. This page exists to make that decision quickly and honestly.
$ ping -c 5 rpc.nln.clr3.orgrtt min/avg/max/mdev = 0.18/0.21/0.27/0.03 ms$ ping -c 5 api.mainnet-beta.solana.comrtt avg 14.9 ms · public internet$ mpstat -P ALL 1 5 | grep -i steal%steal: 0.00 · every core, every sample# pinned cores + private VLAN: what VDS and# bare metal add on top of a VPS.
Own a slice, own the cores, or own the machine
Shared Compute VPS from $100/mo | Deterministic Compute VDS from $245/mo | Dedicated Metal Bare Metal from $458/mo | |
|---|---|---|---|
| Tenancy | Shared host, fair-share vCPU | Pinned physical cores | Entire physical machine |
| CPU steal time | Real, measure with mpstat | 0.00%, contractual | No hypervisor exists |
| RAM ceiling | 64 GB | 128 GB · NUMA-local | 768 GB |
| Storage | 1 NVMe volume | 2–4 NVMe namespaces | 2–4 physical NVMe drives |
| Network | 1 Gbps · 2 TB egress included | 5–10 Gbps committed QoS | 10–100 Gbps unshared |
| Path to NLN fleet | Standard network, measure on first boot | Dedicated private VLAN | Dedicated private VLAN |
| Provisioning | Same-day, most < 1 h | 2–3 minutes | ~24 hours, burn-in tested |
| SLO | 99.99% monthly | 99.99% monthly | 99.99% network & power |
| Failure recovery | Snapshot restore < 15 min | Live migration < 8 min | IPMI + MTTR < 4 h |
| Mainnet validator | No | No (RAM + I/O floor) | Yes, built for it |
| Choose it when | The workload waits on the network | p99 latency has a PnL | You need the hardware itself |
| Explore VPS → | Explore VDS → | Explore Bare Metal → |
Start from the workload, not the price list
Bursts CPU on signals, idles between slots. Network-bound 95% of the time.
The bot spends its life waiting on the wire. 8 shared vCPU covers the duty cycle with room to spare, and the money stays in the strategy.
Step up ifmpstat shows steal > 1% in peak hours, or your loop decodes every transaction.
Hot loop on the gRPC feed, decode everything, race a single thread to the trigger.
Pinned cores keep the decode loop’s p99 glued to its p50: a tail event during your neighbor’s burst is a missed trade.
Step up ifthe race comes down to one thread: 4.4 GHz metal beats 3.0 GHz virtualized.
Constant ingestion from Yellowstone into Postgres on the same box. Never idles.
16 pinned cores and NUMA-local RAM let ingest and Postgres share a box without sharing tails. Unix-socket DB writes, zero network hop.
Step up ifyou index more than a couple of programs or keep > 40 GB hot in Postgres.
Receive events, transform, fan out over HTTP. Spiky concurrency, light CPU.
16 vCPU absorbs concurrency spikes; shared vCPU is irrelevant when the work is mostly I/O.
Step up ifconcurrency is modest, same architecture, $160 less.
Your own read-only RPC node, ledger and accounts on separate volumes.
32 pinned cores, 128 GB, and four NVMe namespaces handle a non-voting mirror, virtualized, so you can resize as it grows.
Step up ifthe mirror is permanent: $203/mo cheaper with physical drives and no hypervisor.
Learning the operator workflow: vote, restart, snapshot, monitor.
16 vCPU / 64 GB carries a testnet validator. The point is the runbook practice, not the hardware.
Step up ifyou want the exact I/O behavior you will see on mainnet.
Voting on mainnet. Skipped slots cost stake. Hypervisor jitter is disqualifying.
24 cores / 384 GB / four drives is the field-standard build. The vote pipeline never crosses a hypervisor.
Step up ifyou are running Firedancer or want 100 Gbps headroom.
Custom Geyser plugin feeding your own downstream: full-chain firehose in, structured data out.
Full-chain Geyser is a sustained-everything workload: real drives for the sink, real cores for the decode, 128 GB to breathe.
Step up ifthe transform is compute-bound: same box at 4.5 GHz.
The next validator client: tile architecture, AF_XDP, enormous memory appetite.
Built for it: 32 cores, 768 GB, 100 Gbps, and we ship a Firedancer-tuned image (governor, IRQ affinity, AF_XDP) on request.
Configure across the whole catalog
The smallest box that runs a real Solana workload.
- ·Own kernel, own IP, full root
- ·Fair-share vCPU and network scheduling
- ·Local NVMe storage, nightly snapshots
- ·Key-only SSH from first boot, any OS
- ·Provisioning: Same-day · most within the hour (EU business hours)
- ·SLO: 99.99% monthly · automatic credits
Not for: Sustained CPU above ~60%; move to VDS before tail latency bites.
The production default for trading bots that wait on the network.
- ·Own kernel, own IP, full root
- ·Fair-share vCPU and network scheduling
- ·Local NVMe storage, nightly snapshots
- ·Key-only SSH from first boot, any OS
- ·Provisioning: Same-day · most within the hour (EU business hours)
- ·SLO: 99.99% monthly · automatic credits
Not for: Hot decode loops at full gRPC volume: that is vds.standard territory.
Headroom for fanout, simulation, and testnet validator work.
- ·Own kernel, own IP, full root
- ·Fair-share vCPU and network scheduling
- ·Local NVMe storage, nightly snapshots
- ·Key-only SSH from first boot, any OS
- ·Provisioning: Same-day · most within the hour (EU business hours)
- ·SLO: 99.99% monthly · automatic credits
Not for: Anything where p99 has a PnL. Shared silicon has shared tails.
Cross-shopFor $105 less, vds.standard trades 8 shared vCPU for 8 pinned cores.
Eight cores nobody else can touch. Steal time: 0.00% by contract.
- ·Physical cores pinned, single socket, no shared hyperthreads
- ·RAM allocated from the same NUMA node as your cores
- ·Committed bandwidth in the QoS class, not borrowed
- ·Dedicated NVMe namespaces, verifiable with lscpu / numactl / mpstat
- ·Provisioning: 2–3 minutes · capacity is held hot
- ·SLO: 99.99% monthly · host-failure migration target < 8 min
Not for: Bursty, mostly-idle workloads: a vps.pro does that for $55 less.
The indexer workhorse: ingest, decode, and Postgres on one NUMA node.
- ·Physical cores pinned, single socket, no shared hyperthreads
- ·RAM allocated from the same NUMA node as your cores
- ·Committed bandwidth in the QoS class, not borrowed
- ·Dedicated NVMe namespaces, verifiable with lscpu / numactl / mpstat
- ·Provisioning: 2–3 minutes · capacity is held hot
- ·SLO: 99.99% monthly · host-failure migration target < 8 min
Not for: Mainnet validators. The RAM ceiling and shared drives rule it out.
The top of virtualization. Past this point, buy the machine.
- ·Physical cores pinned, single socket, no shared hyperthreads
- ·RAM allocated from the same NUMA node as your cores
- ·Committed bandwidth in the QoS class, not borrowed
- ·Dedicated NVMe namespaces, verifiable with lscpu / numactl / mpstat
- ·Provisioning: 2–3 minutes · capacity is held hot
- ·SLO: 99.99% monthly · host-failure migration target < 8 min
Not for: If you need >128 GB RAM, nln.metal.base is $203/mo cheaper with real drives.
Cross-shopnln.metal.base: same 16-core class, 128 GB, 4 physical drives, no hypervisor at all.
Entry metal: high clocks, your own NIC, no hypervisor anywhere.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
Not for: Core-hungry parallel decode: six cores is six cores.
The highest clocks in the fleet. Built for single-thread race paths.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
Not for: Wide parallel workloads: clock beats width here, not the reverse.
Four physical drives: the layout an RPC node actually wants.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
Not for: Mainnet voting; 128 GB is below the field standard.
base, with the clock turned up 50% for compute-bound pipelines.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
The mainnet validator build: field-standard spec, turnkey.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
Firedancer-class: 768 GB, 100 Gbps, sized for the next client.
- ·Full control of IRQ affinity, CPU governor, and kernel
- ·Physical NVMe drives; ledger and accounts DB never share IOPS
- ·10–100 Gbps unshared host networking
- ·IPMI / KVM-over-IP on a dedicated management VLAN
- ·Provisioning: ~24 hours · OS installed, drives validated, burn-in tested
- ·SLO: 99.99% network & power · hardware MTTR < 4 h (business window)
What the private VLAN buys you
One SLA culture across the fleet
- ·VPS pools run fair-share vCPU scheduling with headroom held in reserve
- ·VDS and metal capacity is never shared; pinned capacity is held hot
- ·Spare capacity stays racked and powered, so orders never wait for a host to be carved
- ·AMD EPYC platforms, DDR4 ECC across the fleet
- ·Datacenter-grade storage with redundancy (mdraid1) on virtualized pools
- ·Every metal box burn-in tested before handover: CPU, RAM, drive surface, NIC
- ·VPS: nightly snapshots, restore to a healthy host < 15 min
- ·VDS: live migration off a failing host, target < 8 min
- ·Metal: IPMI / KVM-over-IP always on, hardware MTTR < 4 h in the business window
- ·Committed QoS bandwidth on VDS; 10–100 Gbps dedicated ports on metal
- ·Dedicated private VLAN to the NLN RPC / gRPC / WS fleet included on every VDS and bare-metal server
- ·Fair-share scheduler on shared uplinks; one tenant cannot saturate a host
- ·Full root. No locked kernels, no restricted outbound ports, no inspection
- ·Key-only SSH from first boot; password auth ships disabled
- ·Bring your own ISO: Ubuntu, Debian, Rocky, Arch, NixOS all routine
- ·99.99% monthly rolling, fleet-wide
- ·Credits applied automatically on breach, no ticket required
- ·Only enforcement on your box: outbound SMTP abuse limits
Frequently asked questions
Related products
Shared compute from $100/mo for bots, webhooks, and indexers.
Deterministic compute from $245/mo. Pinned cores, 0.00% steal time.
The whole machine from $458/mo. Validator- and Firedancer-grade.
The endpoint every server on this page reaches in 0.21 ms.
The streaming layer trade-adjacent workloads consume.
Managed plugin slots on production validators.
Pick a tier and deploy today
VPS from $100/mo, VDS from $245/mo, bare metal from $458/mo. One dashboard, one bill, one support queue. VDS provisions in minutes; most VPS orders ship within the hour.
Not sure between two SKUs? Send the workload description and an engineer replies with a sizing call, usually same day.
- 01Create an accountEmail and a password. No card required to look around.
- 02Pick tier + paste your SSH keyThe exact specs and prices on this page. No checkout surprises.
- 03Root SSH lands in your inboxKey-only auth, your chosen OS image, ready for the NLN data fleet.