Isolated and compliant
01Single-tenant clusters with no shared traffic, no shared cache, no shared queue. SOC 2 Type II controls and audit-ready logging on every component.
Dedicated Clusters
Dedicated clusters, configured to your workload.
Single-tenant infrastructure with private networking, custom hardware, and SLAs negotiated against the contract. Built for teams whose workloads will not share tenancy with another customer's traffic, cache, or queue.
100%
Single-tenant
Hardware, network, cache
Custom
SLA terms
Negotiated per contract
12
Regions available
Deploy where you operate
Hours
Provisioning time
Standard profiles
Endless configuration, fully managed
Customized your way. Operated by us.
Every capability below applies to every cluster, regardless of network. The contract scopes capacity, hardware, region, and SLA. The operations team stays the same.
Custom tracers and binaries
02Run instrumented clients, custom tracers, MEV-specific patches, and protocol-internal builds on dedicated hardware. You specify the binary. We operate the deployment.
Any region, any latency target
03Deploy in any of 12 regions worldwide. Co-locate with your matching engine, oracle, or settlement layer for sub-millisecond network latency to chain reads.
Hardware sized to your traffic
04CPU, memory, storage, and NIC profiles selected and benchmarked against your real traffic. No oversized templates. No undersized surprises.
Private networking
05VPC peering, private endpoints, and dedicated transit. No public ingress required for production reads. IP allowlists and mTLS at the gateway.
Capacity-based pricing
06Contracts scope monthly RPS, archive storage, trace volume, and bandwidth. No surprise overages. No auction pricing during volatility windows.
Reliability at every layer
Reliability written into the architecture.
Single-tenant does not mean single-point-of-failure. Every cluster runs redundant by design, fails over automatically, and recovers without operator input.
Redundancy by design
01Multiple full nodes per chain per region with active health checks. Failures route around the affected node within milliseconds, not seconds.
Automatic failover
02Cluster-level failure spills to dedicated standby capacity, then to shared infrastructure if you opt in. Requests stay in flight; nothing drops at the gateway.
Block-perfect consistency
03Nodes within a region stay synced down to the block. Reorg-aware routing prevents cross-node inconsistency on read paths during chain reorganizations.
Real-time observability
04Per-method latency, p99 tails, sync status, and node health on a live dashboard. Grafana endpoints available for your existing observability stack.
Every network, ready
05Active relationships with protocol foundations across 100+ networks. New chains spin up on dedicated capacity within days of mainnet launch.
Live in hours, not weeks
06Standard cluster profiles provision in hours. Custom hardware orders ship within days. URL swap is the migration.
How a cluster ships
From scoping call to production traffic.
Five stages from first benchmark to live cluster, each owned by a named protocol engineer who stays with the account.
- 01
Scoping and workload modeling
We benchmark your real traffic profile across method mix, RPS, archive volume, and latency targets. The cluster spec comes out of measurements, not templates.
- 02
Hardware selection
CPU class, memory profile, NVMe vs SSD, and NIC speed selected against the benchmark. Reservation orders ship within days. Standard profiles available within hours.
- 03
Network and region placement
Deploy in one or more of 12 regions, with VPC peering or dedicated transit to your environment. Cross-region replication and failover configured per contract.
- 04
Custom client deployment
Reference clients, custom tracers, instrumented builds, or protocol-internal binaries deployed and operated by named engineers. CI hooks supported.
- 05
Operations and incident response
Named protocol engineers on a 24/7 rotation per cluster. Sev-1 response in under 15 minutes. Post-mortems publish to your team within 48 hours.
Workloads that need isolation
What teams run on dedicated clusters.
Six workload categories drive most dedicated cluster contracts. Different constraints, same operational posture.
Security and forensics
Custom tracers, full archive access, and reorg replay on dedicated hardware. The trace data flow you depend on for threat detection does not share a cache with anyone else's queries.
Financial institutions
Single-tenant infrastructure with documented controls, SOC 2 Type II posture, and contracted SLA. Your compliance team gets the audit log they need.
Trading and DeFi infrastructure
Co-located clusters with sub-millisecond reads from your matching engine. Pending transaction streams and mempool access on dedicated bandwidth.
Oracles and price feeds
Predictable read latency for high-frequency price publishing. No throttling, no queue depth from another customer's load test.
Analytics and indexing
Archive and trace access without operating archive hardware in-house. Backfill rates that do not compete with production reads.
Centralized exchanges
Production capacity that absorbs deposit and withdrawal storms. Failover paths your risk team will underwrite.
Shared vs dedicated vs in-house
What you trade away at each tier.
Shared infrastructure trades isolation for convenience. Dedicated clusters trade per-request pricing for predictability and control. In-house trades all of it for engineering headcount.
| Metric | Shared infrastructure | Dedicated cluster | In-house operator |
|---|---|---|---|
| Tenancy | Multi-tenant | Single-tenant, isolated network | Single-tenant |
| Performance profile | Burstable, variable under load | Reserved and sized to traffic | Variable, team-dependent |
| Compliance posture | Standard provider controls | SOC 2 Type II, audit-ready | Build your own |
| Custom binaries | No | Yes, supported | Yes, on your team |
| Time to live | Minutes | Hours to days | Months |
| Hardware control | None | Specified to your traffic | Full, on your team |
| Engineering load on your team | Minimal | Minimal | High |
| Cost model | Per request | Reserved capacity contract | Capex plus payroll |
Cluster specs
The contract, in one panel.
Standard inclusions on every dedicated cluster contract. Per-cluster terms scoped during onboarding and written into the master services agreement.
- Single-tenant hardware, network, and cache
- Custom binaries, tracers, and instrumented clients supported
- VPC peering and private transit to your environment
- Named protocol engineer on a 24/7 rotation
- Audit-ready logging and SOC 2 Type II posture
Specs
- Tenancy
- Single-tenant, isolated
- Networking
- VPC peering · Private transit
- Hardware
- Sized to your traffic
- Custom binaries
- Supported
- Regions
- Any of 12 worldwide
- Compliance
- SOC 2 Type II controls
- SLA
- Custom, contracted
- Support
- Named protocol engineer
Related infrastructure
What runs on dedicated capacity.
Frequently asked
Dedicated clusters, answered.
Don't see your question? A named engineer replies within one business day.
Use a dedicated cluster when your workload requires single-tenant performance, custom hardware, custom binaries, or contractual SLAs. Common triggers include compliance requirements, sub-100 millisecond latency targets, archive or trace workloads at high volume, and traffic profiles that exceed shared-pool burst limits.
Security and forensics platforms, financial institutions, trading and DeFi infrastructure, oracle networks, analytics providers, and centralized exchanges. Any workload where reliability, performance, and compliance posture are contractual rather than aspirational.
Yes. Most customers run dedicated clusters for production traffic and shared pools for development and staging. Failover from dedicated to shared is supported and configured per endpoint.
Yes. We operate the hardware, software, monitoring, and incident response. You consume endpoints and pay against the capacity contract. Named protocol engineers cover every cluster on a 24/7 rotation.
Standard cluster profiles provision within hours of contract sign-off. Custom hardware configurations ship within days. Network and region selection happens during the initial scoping call.
Burst capacity is configurable per cluster. Overflow either spills to dedicated standby capacity or to shared infrastructure at your election. Hard caps available for cost-controlled deployments.
Dedicated clusters price against reserved capacity, not per-request volume. Contracts scope monthly RPS, archive storage, trace volume, and bandwidth. No surprise overages. No auction pricing during volatility windows.
In-house operation requires hiring SREs, building HSM integrations, sourcing hardware, and maintaining 24/7 coverage across every protocol you support. Dedicated clusters deliver the same isolation and performance without the headcount load. You retain configuration authority; we operate the deployment.
Yes. Validator workloads run on dedicated clusters with HSM-backed signing, multi-region failover, and full slashing-protection databases. Custom validator clients, instrumented builds, and protocol-internal patches are supported and operated by named protocol engineers.
Put your workload on infrastructure that's built to run it.
One conversation. A named engineer. A reply within one business day. No custody transfer. No lock-in.