Skip to main content
Every Boost response identifies who answered, how long Boost spent on the request, and how many request items were served without reaching your provider.

​
Response headers

The x-edge-billable name predates Boost’s free pricing. Treat it as a served-item count, not a monetary charge.

​
Single requests

For a single JSON-RPC request:
  • x-cache: HIT and x-edge-billable: 1 mean Boost served the call without sending it upstream.
  • x-cache: MISS and x-edge-billable: 0 mean the configured endpoint answered.
  • x-edge-source: static identifies methods answered directly by the Boost edge, such as eth_chainId.

​
Batch requests

Read x-edge-billable instead of relying on x-cache alone. x-cache is HIT only if Boost served every item. For a batch of 45 calls where Boost serves 44:
The batch contains one forwarded item, so the aggregate cache header is a miss even though most calls avoided the upstream.

​
Calculate an avoided-upstream rate

Count JSON-RPC items, not HTTP requests:
This rate measures calls Boost served. To estimate monetary savings, apply your provider’s pricing model to those avoided calls. Providers can price methods, compute units, or plans differently, so the headers do not represent an exact dollar amount.

​
CLI metrics

Fetch aggregate endpoint metrics with:
The command reports cache hits, latency, miss reasons, and provider-level statistics. See the goldsky boost metrics reference for its current output and options.

​
What to monitor

  • Served-item rate: The share of JSON-RPC items with x-edge-billable greater than zero
  • Miss reasons: Whether calls miss because of method eligibility, block tags, unavailable data, or batch size
  • Boost duration: x-edge-duration-ms, split by response source
  • Upstream errors and latency: Requests with x-edge-source: endpoint, grouped by provider
If an eligible method repeatedly misses, use the troubleshooting guide.