What We Test
We measure how quickly each provider lands a Solana transaction on-chain from submission to confirmed slot. “Landing” means the transaction appears in a confirmed block. We do not test MEV protection, bundle landing, or websocket latency.
Infrastructure
Location: EU West (Frankfurt). All providers are tested simultaneously — identical transactions are sent to all endpoints at the same instant within each run.
Transaction Parameters
Each run uses an identical base transaction sent to all providers. Priority fees are equal for all providers within each run. Tips (Jito-style) are added only where the provider supports them — currently Jito and Nozomi — at each provider's standard tip amount.
Latency Measurement
Reported latency is not raw wall-clock time. From the send-to-observe interval we subtract the estimated network transit time on both ends: the round-trip to the provider endpoint on the way out, and the approximate shred propagation time from the leader back to our monitoring node. This isolates the provider's own relay and inclusion speed rather than geographic distance.
Landing confirmation uses shred-resolution detection: we monitor the Turbine shred stream directly and timestamp the moment the first shred carrying the transaction arrives — giving single-shred precision instead of the coarser polling-based approach.
Metrics
- Landing Rate: percentage of submitted transactions confirmed on-chain within the measurement window.
- p50 / p95 / p99 (ms): corrected latency (see above) at the 50th, 95th, and 99th percentile across all transactions in the measurement period.
- Slot distribution (+0, +1, +2…): number of Solana slots elapsed between submission and confirmation. +0 means confirmed in the immediate next produced slot. Slot distance is immune to clock and geographic effects and is consistent across all measurement locations.
Verification
Every individual result includes a link to Solana Explorer (solscan.io) so anyone can independently verify the on-chain transaction.
Caveats
- Results reflect EU West measurements. Performance from other regions may differ significantly.
- Solana network load varies. High congestion affects all providers simultaneously.
- Provider configurations change. Historical data reflects conditions at the time of measurement.
- The current block leader identity influences slot distribution results.
- Shred propagation correction is approximate — small residual geographic effects may remain.
- Tips differ between providers where applicable, which may influence inclusion priority in addition to raw relay speed.
Independence
Operated by Netticode Oy. We test our own RPC alongside competitors. No commercial agreements with any provider influence results. All providers run under identical base conditions; tip differences reflect each provider's own requirements.