Scenario
Round-trip latencies for chained requests getSlot and getBlock(slot) with start-to-start throttling of (at least) 1100 ms for every request.
Also, slot number consistency with block availability and block age. The block propagation comparison is conservative and latency-agnostic.
Of the possible 31 regions, 2 were selected for data collection: Columbus, OH and Frankfurt.
Each location probe breaks its processing loop if / when the endpoint response is too-many-requests, a.k.a. HTTP code 429 etc. Also, codes 401 and 403.
At the time of the report generation: 4 of 4 datasets were available from 2 selected regions x 2 endpoints.
Findings
| Endpoint | Lagging | Unique | Lowest | Arrival Tms | Arrival Region | Highest | Arrival Tms | Arrival Region |
|---|---|---|---|---|---|---|---|---|
| A. edge.Erpc.global | 0 | 16 | 415,893,352 | 1,777,250,033,644 | Frankfurt / AWS | 415,893,430 | 1,777,250,064,482 | Frankfurt / AWS |
| B. solana.api.Pocket.network | 10 | 15 | 415,893,352 | 1,777,250,033,659 | Frankfurt / AWS | 415,893,434 | 1,777,250,066,373 | Columbus, OH / AWS |
We detected 10 responses whose sendTms, arrivalTms and number were such that we can state the endpoint was lagging in the respective region.
E.g. on arrivalTms=1,777,250,042,531 in Columbus, OH / AWS, we received number 415,893,374 from A. edge.Erpc.global. Afterwards, on sendTms=1,777,250,042,560 we dispatched a request to B. solana.api.Pocket.network, which returned lower number 415,893,373 at arrivalTms=1,777,250,042,716. Therefore, B. solana.api.Pocket.network was lagging because it was returning lower number 415,893,373 well after A. edge.Erpc.global advanced to 415,893,374.
Details
Latencies exclude one warmup request per endpoint in each region. Additional columns are visible when you scroll to the right.
Errors
Info about errors incl. 429. Up to 4 retained error responses per endpoint, per region. Scenario-derived availability notes may also appear.