Control
01- Self-hosted
- Full OS and network control
- Managed provider
- Agreed service configuration
Blockchain infrastructure
Blockchain infrastructure
for Web3 and crypto products
Private nodes, RPC endpoints, monitoring, security and scaling for Bitcoin, Ethereum, TON, Solana, Polygon, BNB Chain and other networks — a managed access layer instead of a separate platform team for every protocol.
- 4
- Networks in production
- 16 ms
- Median response
- 99.99%
- Endpoint uptime
- 9
- Regions
The stack behind one address
One endpoint, four layers of operations
- Your applicationyour side
timeouts, retries, reorgs
- rpc.settla.net/sol/mainnet
- operated by Settla
- Endpoint
key, limits, environments
- Routing
pool health, rotation
- Nodes
full and archive, upgrades
- Endpoint
- Solana network302,847,119
a slot every 0.4 s
Access protocols · HTTP · WS · gRPC
Components
Blockchain infrastructure components
Blockchain infrastructure includes more than a node server. Production systems need endpoints, monitoring, access control, upgrades, capacity planning and failure handling.
Nodes
Full or archive clients for supported mainnets and testnets.
full · archive
Endpoints
HTTP, WebSocket and network-specific interfaces for applications.
HTTP · WebSocket · gRPC
Monitoring
Head lag, method checks, errors, latency and resources.
head lag · p95 · error rate
Operations
Updates, capacity planning, incident response and recovery.
upgrades · capacity · on-call
The boundary runs through the endpoint
Settla operates
- Node sync and health
- Client software upgrades
- Rotation and draining
- Capacity planning
endpoint
Your application handles
- Timeouts and retries
- Rate limit handling
- Chain reorganisations
- Idempotent operations
Monitoring and security
A node returns 200 while lagging behind the chain
An availability check sees a live process, an open port and a valid response. It does not see lag at all — worse, a stalled node answers faster than a healthy one, because it has nothing to catch up on. Switch the check and watch what the client receives meanwhile.
both nodes in sync
Pool behind /sol/mainnetSolana · a slot every 0.4 s
- sol-fra-01eu-centralHTTP 200302,847,181in sync
- sol-ams-01eu-westHTTP 200302,847,181in sync
The client read stale state for 16 s
- network blocks
- no new blocks
- 26 s window
- 0
- Node lag, slots
- current
- Client state age
- 0%
- Requests to the stale node
- 2 / 2
- Nodes in rotation
Check height and finality
Local height is compared against an independent source, not against itself. Alongside it go control RPC methods, error rate, p95/p99 and free resources.
A threshold, not the first miss
One second behind the chain is a hiccup, not an outage. A node leaves rotation on a threshold: 15 slots, roughly 6 s on this network.
An alert points to a runbook
A notification without a runbook only records the problem. On-call needs the next step, not a description of what happened.
Access control is the other half of the job
- Credentials per environment
- Separate keys for production, staging and local development, with rotation and usage audit.
- Least privilege
- IP and domain restrictions where they are supported, plus protected secret storage.
- Wallet keys stay with you
- Private keys are never handed to a node provider without a separately justified custody architecture.
Scaling
Scaling Web3 workloads
Scaling starts with splitting the load, not with adding nodes: indexers, user reads and transaction writes must not compete for one quota.
One quota for everything
A backfill drains the limit, and a transaction waits in line behind it.
- User reads
- Transaction writes
- Indexers and backfill
Lanes separated
Each lane gets its own limit, its own cache and its own queue.
- User readscache safe reads
- Transaction writesreserved quota
- Indexers and backfillqueue and cap heavy methods
User reads
Balances, statuses and history in the interface. Numerous, small and mostly safe to cache.
eth_call · getBalance
Transaction writes
A rare but critical lane — the first one to suffer when someone else drains the quota.
eth_sendRawTransaction
Indexers and backfill
Background work with heavy methods. It needs volume, not milliseconds.
eth_getLogs · trace_block
Several nodes and health-aware routing
Growth adds nodes, and routing accounts for their health — the health measured by block height rather than by HTTP status.
A reorg is not an application error
A block you already read may stop being part of the chain. Finality differs between networks, so retries have to be idempotent.
Choosing a model
Self-hosted vs managed infrastructure
Both models work. The difference is not reliability but who carries the on-call rota and what the next network costs to launch.
| Criterion | Self-hosted | Managed provider |
|---|---|---|
| Control | Full OS and network control | Agreed service configuration |
| Operations | Customer team, around the clock | Provider handles core node operations |
| New networks | A new deployment project | Connect an available endpoint |
| Cost | Servers, engineers, on-call, redundancy | A plan or a custom configuration |
| Best fit | Special requirements and a mature platform team | Product focus and faster multichain access |
Operations
02- Self-hosted
- Customer team, around the clock
- Managed provider
- Provider handles core node operations
New networks
03- Self-hosted
- A new deployment project
- Managed provider
- Connect an available endpoint
Cost
04- Self-hosted
- Servers, engineers, on-call, redundancy
- Managed provider
- A plan or a custom configuration
Best fit
05- Self-hosted
- Special requirements and a mature platform team
- Managed provider
- Product focus and faster multichain access
Blockchain infrastructure for business
Blockchain infrastructure for business is not a one-time setup but an operational layer that performs reliably around the clock. Companies integrating blockchain into payment flows, custodial services or compliance pipelines need predictable infrastructure.
- Agreed limits
- Isolated endpoints per environment and request ceilings known in advance, not discovered during an outage.
- Clear upgrade procedures
- Client upgrades and network hard forks follow a plan instead of arriving as a Friday surprise.
- Documented incident protocols
- Lag monitoring, response processes and post-incident reviews instead of a chat thread.
Use cases
Use cases for Web3 products
The same access layer serves products with very different load — from a wallet making a dozen calls per screen to an indexer reading history for months.
Wallets and payment products
Several networks in one interface: balances, transfers, statuses and credit notifications.
Learn moreCEX/DEX and deposit systems
Deposit and withdrawal monitoring, where a data mismatch costs money directly.
Learn moreDeFi, NFT, gaming and smart contracts
Reading state, calling contracts and broadcasting transactions from the application.
Learn moreExplorers, indexers and analytics
Full history of blocks, logs and traces for dashboards and investigations.
Learn moreCompliance and risk analytics
Address screening, flow tracing and corporate reporting.
Learn more
Getting started
Blockchain infrastructure for Web3 applications
Start from business-critical operations and application SLOs: which data is read, how quickly transactions must be sent and what event loss is acceptable.
A provider handles node operations, but responsibility remains shared: the application must handle timeouts, rate limits, reorgs and retries correctly. This approach produces more resilient crypto infrastructure than relying on a single supposedly infallible endpoint.
Name the critical operations
What actually breaks for the user when an answer arrives a second late — or never arrives.
Fix the networks and methods
Networks, environments and methods including the heavy ones: both limits and cost follow from that list.
Decide on depth
Whether you need archive and tracing. That is a separate class of nodes, not a checkbox in a plan.
Describe failure behaviour
Timeouts, retries, fallback routing and reorg handling live in application code.
FAQ
Blockchain infrastructure FAQ
Infrastructure shouldn't be the thing slowing you down.
Free to start. No credit card required.