Almost all first detections went to Fastlane
Shreder Fastlane delivered the first valid update for 99.94% of matched transactions in the published sample.
// Geyser gRPC benchmark
A head-to-head benchmark comparing Shreder Fastlane and Shyft Geyser gRPC for Solana Geyser gRPC delivery in Frankfurt.
This page uses the Shyft Geyser gRPC benchmark data only. It does not use Shyft RabbitStream data from the separate shred/Binary benchmark run.
// Published run
The benchmark hub contains two Shyft-related runs. This page expands only the Geyser gRPC run: Shreder Fastlane vs Shyft Geyser gRPC, measured in Frankfurt.
| Benchmark | Shreder Fastlane vs Shyft Geyser gRPC |
|---|---|
| Product class | Yellowstone-compatible Solana Geyser gRPC |
| Shreder endpoint | Shreder Fastlane Frankfurt |
| Comparison endpoint | Shyft Frankfurt |
| Test location | Frankfurt, Teraswitch FRA2 |
| Sample size | 10,000 transactions |
| RTT min / avg / max / mdev | 0.078 / 0.096 / 0.127 / 0.015 ms |
| Benchmark tool | GeyserBench |
// First delivery
The first-detection share records which endpoint delivered the first valid update for each matched transaction seen by both providers.
| Provider | First-detection share | p50 latency delta | p95 latency delta | p99 latency delta |
|---|---|---|---|---|
| Shreder Fastlane Frankfurt | 99.94% | 0.00 ms | 0.00 ms | 0.00 ms |
| Shyft Frankfurt | 0.06% | 14.03 ms | 59.51 ms | 79.89 ms |
A 0.06% first-detection share does not mean Shyft missed the other transactions. It means Shyft delivered the first valid update for 0.06% of matched transactions in this run.
// Reading the result
This is a Geyser-specific benchmark, so the useful question is time-to-data through comparable Yellowstone-compatible streams.
Shreder Fastlane delivered the first valid update for 99.94% of matched transactions in the published sample.
Shyft’s p95 and p99 latency deltas were 59.51 ms and 79.89 ms in this run.
Use this run as a reference, then compare both endpoints with your production accounts, programs, commitment level, and region.
// Provider comparison
Both streams are evaluated as Geyser-style delivery paths. For production decisions, rerun the comparison using the filters, region, and commitment level that match your workload.
| Category | Shyft Geyser gRPC | Shreder Fastlane |
|---|---|---|
| Interface | Geyser gRPC | Yellowstone-compatible Geyser gRPC |
| Benchmark category | Post-execution Geyser stream | Post-execution Geyser stream |
| First-detection share | 0.06% in this run | 99.94% in this run |
| Latency deltas | p50 14.03 ms · p95 59.51 ms · p99 79.89 ms | Baseline row in this run |
| Best evaluated by | Your own endpoint, region, and filters | Your own endpoint, region, and filters |
// Run your own test
Run both Geyser endpoints from the same machine and region, then compare the same filters over the same measurement window.
Use the same benchmark machine, same region, and same measurement window.
Use the same account or transaction filters, commitment level, and timestamping method.
Compare first-detection share, p50, p95, p99 deltas, missed updates, duplicates, reconnects, and stream errors.
This page compares Shreder Fastlane with Shyft Geyser gRPC. It does not use the separate Shyft RabbitStream benchmark data from the shred/Binary section.
In this published Frankfurt run, Shreder Fastlane delivered 99.94% of matched transactions first, while Shyft Frankfurt delivered 0.06% first. The published Shyft latency deltas were p50 14.03 ms, p95 59.51 ms, and p99 79.89 ms.
No. First-detection share only records which endpoint arrived first among matched transactions observed by both endpoints. It does not mean the comparison endpoint failed to deliver the other transactions.
Yes. Fastlane keeps a Yellowstone-compatible Geyser gRPC interface, so compatible clients generally keep the same client model while changing endpoint and token.
Shreder Fastlane can be used by teams evaluating alternatives to Shyft Geyser gRPC when they need Yellowstone-compatible Solana streaming with lower latency. The best way to compare both providers is to benchmark them from the same machine, region, commitment level, and filters.
Run both endpoints from the same benchmark machine, region, commitment level, filters, measurement window, and timestamping method. Compare first-detection share, p50, p95, p99 latency deltas, missed updates, duplicate updates, reconnects, and stream errors.
// Test your own Geyser stream
Use this published Frankfurt run as a reference, then test the filters, region, and commitment level that match your production workload.