Methodology
Network benchmark methodology
How intra-region TCP throughput and round-trip latency between cloud instances, and latency to major Turkish and European cities, are measured with iperf3 and mtr.
Version 1.0 Updated:
Scope
- TCP throughput between two instances in the same region over private networking
- Round-trip latency between instances in the same region
- Round-trip latency from the instance to public measurement endpoints in İstanbul, Ankara, Frankfurt and Amsterdam
Topology
Two identical instances (the 4 vCPU shape from the compute methodology) are created in the same region and, where the provider supports it, the same availability zone, connected by the provider’s private network.
Throughput
iperf3 -c <peer> -P 8 -t 60 -J measures aggregate TCP throughput with 8 parallel streams. The test runs in both directions. Published instance bandwidth limits are recorded next to the result, because many providers cap bandwidth per instance size.
Latency
mtr --report --report-cycles 300 <target> records round-trip time to the peer instance and to external targets. The median and p95 RTT are reported. External targets are fixed public endpoints listed with each result.
Repetitions and verification
Throughput and intra-region latency are measured 5 times at different hours across two days. External latency is measured hourly over 24 hours. Verification requires a second run within ±10% for throughput and ±1 ms (or ±10%, whichever is larger) for latency.
Limitations
- Internet latency depends on routing at the time of testing and on the vantage point.
- Egress pricing is not part of this benchmark; see the pricing methodology.