indexing-performance

name: indexing-performance description: >- Use when optimizing indexer speed or tuning sync performance. HyperSync vs RPC, batch size, RPC tuning parameters, WebSocket config, and preload optimization.

Performance Tuning

HyperSync (Default — Fastest)

HyperSync is the default data source for supported chains. Up to 1000x faster than RPC. No configuration needed — automatic for supported networks.

Batch Size

full_batch_size: 5000  # Target events per batch (default: 5000)

Reduce if handlers make many slow Effect API calls that can't be batched.

RPC Tuning

When using RPC as a data source, tune these parameters per chain:

chains:
  - id: 1
    rpc:
      - url: ${RPC_URL}
        for: sync                    # sync | live | fallback
        ws: ${WS_URL}               # WebSocket for lower-latency live block detection
        initial_block_interval: 5000 # Starting blocks per query
        backoff_multiplicative: 0.8  # Scale factor after RPC error (0.5-0.9)
        acceleration_additive: 1000  # Blocks added per successful query
        interval_ceiling: 10000      # Max blocks per query
        backoff_millis: 5000         # Wait after error before retry (ms)
        query_timeout_millis: 20000  # Cancel RPC request after this (ms)
        fallback_stall_timeout: 5000 # Switch to next RPC after stall (ms)
        polling_interval: 1000       # Check for new blocks every N ms (default: 1000)

RPC `for` Options

Value	Description
`sync`	RPC as main data source for both historical and live
`live`	HyperSync for historical, switch to RPC for live (lower latency)
`fallback`	Backup when primary stalls (default when HyperSync available)

WebSocket for Live Indexing

Add ws: for lower-latency new block detection via eth_subscribe("newHeads"):

rpc:
  - url: ${ENVIO_RPC_ENDPOINT}
    ws: ${ENVIO_WS_ENDPOINT}
    for: live

Database Indexes

Add @index to schema fields for faster queries — see indexing-schema for full syntax (single-field, composite, DESC direction).

Deep Documentation

Full reference: https://docs.envio.dev/docs/HyperIndex-LLM/hyperindex-complete