SettleMint
ArchitectureComponentsInfrastructure

EVM RPC Node

The EVM RPC Node provides the direct interface to blockchain networks, handling JSON-RPC requests for transaction submission, state queries, event subscription, and hybrid node access across public and private EVM-compatible chains.

Overview

The EVM RPC Node is the blockchain network access point for DALP operations. It provides transaction submission, state queries, and event streaming through JSON-RPC.

Production deployments connect through the Chain Gateway for load balancing and failover across multiple endpoints.

The Transaction Signer submits signed transactions through this network layer.

The Chain Indexer consumes chain events after they are available.

Network connectivity

Rendering diagram...

Supported networks

DALP operates across all EVM-compatible networks without application changes. Network-specific configuration handles consensus differences, gas models, and confirmation requirements.

Layer 1 mainnets

NetworkDescription
EthereumPrimary EVM network with largest ecosystem
Polygon PoSHigh-throughput sidechain with low fees
BNB Smart ChainEVM-compatible chain with fast finality
Avalanche C-ChainSub-second finality with Snowman consensus
XDC NetworkEnterprise-grade hybrid blockchain
Gnosis ChainCommunity-owned network (formerly xDai)

Layer 2 rollups

NetworkTypeDescription
Arbitrum OneOptimisticLargest L2 by TVL with Nitro technology
BaseOptimisticOP Stack chain incubated by Coinbase
OP MainnetOptimisticBedrock-based flagship OP Stack chain
LineaZKConsenSys zkEVM with type 2 equivalence
zkSync EraZKMatter Labs ZK Stack rollup
ScrollZKCommunity-driven zkEVM
Polygon zkEVMZKPolygon type 2 zkEVM

Specialized and appchains

NetworkDescription
ADI ChainUAE-based ZK Stack chain for institutional finance and regulated stablecoins
Immutable zkEVMGaming-focused Polygon zkEVM
WorldchainWorld ID verified OP Stack chain

Testnets

NetworkMainnet equivalent
SepoliaEthereum testnet
HoleskyEthereum validator testnet
AmoyPolygon PoS testnet
Arbitrum SepoliaArbitrum One testnet
Base SepoliaBase testnet

Private networks

ClientUse case
Hyperledger BesuEnterprise features with permissioning
Go-Ethereum (Geth)Reference implementation for private networks
Nethermind.NET-based client with enterprise plugins
ErigonArchive-optimized client for analytics

Network resources

  • Chainlist - Complete EVM network directory with public RPC endpoints
  • L2Beat - Layer 2 network security and TVL analytics

RPC capabilities

  • Transaction submission: Routes signed transactions to the network mempool with structure validation before broadcast
  • State queries: Read current or historical blockchain state (balances, contract storage, call simulations)
  • Event subscription: WebSocket connections for real-time event streaming and new block notifications
  • Trace APIs: Debug and trace APIs for transaction execution details and gas breakdowns (client-dependent)

Deployment patterns for node access

DALP does not require one fixed blockchain node hosting model. The production pattern is to route DALP services through the Chain Gateway, then configure one or more EVM RPC upstreams behind that gateway.

PatternWhat DALP connects toWhen it fitsAvailability check
Bank-operated nodeAn internal HTTP, HTTPS, WS, or WSS endpoint exposed by the operator's full node or validator-adjacent infrastructureThe operator needs direct control over node software, network policy, or independent verificationMonitor node lag, endpoint health, supported methods, and whether gateway traffic can move to a secondary upstream
Managed RPC providerA provider endpoint configured as a gateway upstreamThe operator prefers provider-managed node operations, scaling, and supportMonitor provider status, rate limits, authentication failures, and fallback provider readiness
Hybrid node accessA mix of bank-operated and provider-managed upstreamsThe operator wants an internal primary path with provider fallback, or a provider primary path with a bank-operated verification nodeTest failover across both upstream types and reconcile chain reads through the indexer after failover
Private EVM networkOne or more RPC endpoints from the permissioned networkThe target chain is operated by the institution, consortium, or private network operatorTrack consensus health, finality depth, RPC method support, and indexer catch-up after network incidents

This keeps node ownership separate from DALP application behavior. DALP centralizes routing, retries, hedging, cache use, and integrity checks in the gateway.

The deployment still owns node patching, endpoint credentials, private links, TLS or mutual TLS policy, and provider contracts.

Cloud provider and regional boundary

For self-hosted deployments, DALP supports AWS, Azure, GCP, OpenShift, and approved Kubernetes environments when the required managed services, network paths, object storage, observability, and backup controls are available. Region selection belongs to the operator's landing zone and regulatory boundary rather than to the EVM RPC Node itself.

Use this boundary when reviewing hybrid architectures:

RequirementPublic DALP answerEvidence to validate before go-live
Core platform on-premises, node access in cloudSupported when the on-premises DALP services can reach the cloud RPC upstreams through the Chain Gateway and the selected private connectivity, TLS, authentication, and routing policy.Gateway upstream configuration, private connectivity test, credential rotation process, and RPC health alerts
Cloud-hosted Chain Indexer or recovery-region indexerDepends on the selected high availability pattern. The indexer region needs PostgreSQL, RPC endpoints, contract metadata, secrets, and observability.Indexer checkpoint state, database access, RPC access, block-lag alerting, and catch-up drill result
AWS, Azure, or GCP region choiceSupported where the operator can provision the required Kubernetes, PostgreSQL, cache, object storage, backup, and observability services.Region approval, managed-service availability, network design, data-residency approval, and provider SLA mapping
Regional cloud outage affecting one node or indexer pathDALP continues on the remaining region only when a healthy RPC upstream, indexer path, database path, and routing runbook are already configured and tested. Otherwise affected workflows fail until the required chain access or indexed state recovers.Failover drill, measured recovery time, accepted RTO/RPO target, indexer catch-up evidence, and post-failover chain-read check

Treat RTO and RPO as deployment targets, not automatic product promises. The high availability pages define the planning ranges: cloud-native targets 2 to 15 minutes RTO with seconds to 1 minute RPO, hot-warm targets 30 to 180 minutes RTO with 5 to 60 minutes RPO, hot-cold targets 8 to 72 hours RTO with 4 to 24 hours RPO, and hot-hot targets 1 to 10 minutes RTO with seconds to minutes or 1 to 5 minutes RPO depending on the operating model. The measured result comes from the operator's restore or failover drill.

Security

  • Authentication: API keys for managed providers; reverse proxy, service mesh, or node-level authentication for self-hosted nodes
  • Network isolation: Private nodes should stay inside isolated network segments with firewall, ingress, or route restrictions
  • Transport protection: Use HTTPS or WSS for external RPC endpoints. Keep HTTP or WS endpoints internal to trusted network paths, or protect them through private connectivity, ingress policy, or service-mesh controls.

Availability and recovery

Node access and indexing recover together during incidents. After an RPC endpoint fails over, operators should confirm that transaction submission still works, chain reads return the expected head, and the Chain Indexer is catching up from its stored checkpoint.

For production deployments, keep at least two reachable upstream endpoints where the network design supports it.

Monitor RPC endpoint availability, chain head age, finality lag, indexer block lag, and gateway errors.

Include those checks in the selected high availability pattern so node failure does not look like a healthy DALP deployment with stale chain data.

See also

On this page