Single-tenant blockchain infrastructure
Isolated blockchain clusters sized to your read/write mix and peak RPS. Predictable latency, elastic throughput, multi-region failover, observability, and custom SLAs.
Dedicated Clusters trade burst-from-pool economics for full isolation. Compute, memory, network, and storage are allocated to your workload. No shared rate limits, no neighbor-induced tail latency.
Tier 1
Many tenants behind one pool. Per-customer rate limits, shared compute and network.
Picked when
Hobby projects, prototypes, low-traffic dApps.
Tier 2
Reserved capacity on shared infrastructure. Better tail latency than pooled, but neighbors still exist.
Picked when
Growing apps that need predictable capacity without full isolation.
Tier 3
Single-tenant compute, memory, network, and storage. No neighbors, so latency moves with your traffic alone.
Picked when
Exchanges, DeFi protocols, MEV, wallets, regulated workloads.
Exchanges
Hot-wallet RPC, withdrawal broadcasting, and deposit indexing on isolated capacity. No competing tenants during volume spikes or market events.
Institutions
Compliance-friendly infrastructure with audit-ready logs, IP allowlists, mTLS, and SSO with your identity provider.
Bots
Dedicated Nodes placed in the region your strategy runs from, with private mempool access on supported chains. Built for latency-bound strategies.
Single-region clusters fail under regional outages. Every workload runs active-active across at least two regions. Automatic failover, client unaware. Latencies below are measured from regional ingress to nearest cluster node.
8 regions · reference topology
Default
Anycast routes requests to the closest healthy region. Default for read-heavy workloads.
Per method
Per-method routing. Broadcast lands at the lowest-latency mempool, reads at the cheapest region.
On incident
Cutover in <500ms on regional incidents, behind the same endpoint URL.
Rollouts
Pin a percentage of traffic to a specific region for canary rollouts or compliance.
An RPC SLA that covers uptime alone leaves out the failures you notice. Ours also sets p95 and p99 latency per method, sustained and burst RPS, error-rate ceilings, and incident response, with automatic credits when we miss.
| Metric | What it measures | Standard | Premium | Remedy |
|---|---|---|---|---|
| Uptime | Monthly availability per endpoint | 99.95% | 99.99% | Auto credits, tiered |
| p95 latency | Per-method 95th percentile RTT to regional edge | Set in your contract | Set in your contract | Method-scoped credit |
| p99 latency | Per-method tail latency, the figure that hurts when it slips | <100ms | <80ms | Method-scoped credit |
| Sustained RPS | Contracted request capacity, never throttled below it | Custom | Custom + burst | Right-size on request |
| Error rate | 5xx ceiling over rolling window | Set in your contract | Set in your contract | Credit + root-cause |
| Incident response | An on-call engineer acknowledges and owns the incident | Set in your contract | Set in your contract | Shared incident bridge |
| Failover | Active-active regional cutover | <500ms | Set in your contract | No client-side retries |
Uptime
p95 latency
p99 latency
Sustained RPS
Error rate
Incident response
Failover
We measure ourselves against the SLA, log every breach, and credit your account automatically. You never file a ticket to claim one.
A global average can hide one slow method. We write latency terms against p95 and p99, scoped per method, so each path is measured on its own.
We share a redacted sample SLA during scoping. You see the exact wording, remedies, and exclusions before you sign.
Per-method latency, error rate, and saturation metrics. Live dashboards, anomaly detection, and integration with your existing observability tools (Grafana, Datadog, PagerDuty).
SOC 2 Type II infrastructure, private networking, IP allowlists, mTLS, and key-vaulted credentials. SSO with your identity provider. Audit-ready logs retained per your policy.
We profile your read/write mix, peak RPS, method usage, and tail-latency targets. Output: a sizing recommendation and a cost model you can compare against your current spend.
Region selection, replica topology, execution-client mix, indexer modules, networking, and observability. Each choice is checked against your per-method p99 target.
Cluster provisioned and tested in parallel to your existing setup. Cutover behind your control. Incremental traffic shifting or a clean switch on your schedule.
24/7 monitoring, incident response, client upgrades, and regular reliability reviews. SLA credits apply automatically. Your architect adjusts capacity on request, without a ticket queue.
Need something specific? Talk to an engineer.
A single-tenant deployment of dedicated nodes. Compute, memory, network, and storage are allocated to your workload alone. No shared neighbors, no shared rate limits, and no competition for resources during traffic spikes. The output is a private RPC Endpoint with capacity, latency, and uptime guarantees written into your contract.
Shared RPC pools many customers behind one cluster with per-customer rate limits. Cost-efficient, but tail latency is sensitive to neighbor behavior. Semi-dedicated reserves capacity but shares the underlying infrastructure. Dedicated clusters isolate compute, memory, network, and storage. Predictable tail latency and contractually guaranteed throughput.
When tail latency matters (liquidations, arbitrage, market data), when you serve a large user base and noisy-neighbor risk is unacceptable, when compliance requires isolated tenancy, or when usage is high enough that dedicated capacity is cost-competitive with bursty shared tiers. Typical inflection point: sustained high request volume, or a p99 requirement at the <80ms dedicated target.
Any of the 28+ chains we run Dedicated Nodes on, including Ethereum, Arbitrum, Base, Polygon, BNB Chain, Avalanche, Cosmos, Polkadot, Tron, and Bitcoin. Mainnet, and testnet where the chain has one. If your chain isn't listed, ask: we confirm a provisioning timeline for a new network after engineering review.
Yes. Active-active deployments across two or more regions with automatic failover in <500ms. Routing strategies include geo-nearest, latency-targeted, and weighted distribution. Failover is transparent to your client. No retries or sticky-session logic required on your side.
Uptime (typically 99.95% to 99.99%), p95 and p99 latency targets per method, sustained and burst RPS, error-rate ceilings, and incident response time. SLAs are written into the contract with automatic service credits, not credits-on-request. We provide a sample SLA before negotiation.
Live dashboards for per-method latency, error rate, and saturation. Anomaly detection on traffic patterns and tail latency. Native integration with Grafana, Datadog, PagerDuty, and Slack. Logs and metrics can be streamed into your existing observability stack via standard protocols (OTLP, Prometheus remote write).
Yes. Specify execution client version (Geth, Erigon, Nethermind, Besu, Reth, etc.), consensus client for Ethereum (Lighthouse, Prysm, Teku, Nimbus), Bor and Heimdall for Polygon, and indexer modules (trace and debug namespaces, Otterscan on Erigon). We deploy and operate to your spec.
Pricing is quoted per engagement, based on cluster size, region count, replica topology, SLA tier, and chain. We deliver a cost model after the capacity assessment so you can compare against your current spend before contracting.
Request a quote →We commit to a provisioning date in the contract, after the capacity assessment. Single-region clusters come up faster than multi-region deployments. Archive nodes with full historical state take longer to sync, but the cluster serves forward traffic as soon as it is provisioned.
Yes. Instanodes is SOC 2 Type II certified, and the report is available under NDA. Our controls cover access, encryption, change management, and incident response.
Yes. Every Dedicated Cluster contract includes a named architect for capacity planning, migrations, and regular business reviews. Your architect is your single point of contact across products.
Yes. We run your Dedicated Cluster in parallel to your existing provider, then shift traffic in steps you set, for example 1%, 10%, 50%, then 100%, over a window you control. You can roll back at any step, and your clients see no downtime.
Share your traffic profile, latency targets, and chain mix. We send back a sizing recommendation, a sample SLA, and a price, before contract.