Throughput & speed¶
My transfers are slower than my tier — why?¶
The most common cause is a single-stream transfer, which can't saturate a high-bandwidth link on its own. Run parallel transfers and your aggregate throughput jumps. See Performance.
How do I run parallel transfers?¶
rclone copy ./data zendrive:my-bucket/ --transfers 16 --checkers 16
# s3 sync parallelizes automatically; tune concurrency:
aws configure set s3.max_concurrent_requests 20 --profile zendrive
aws --profile zendrive --endpoint-url "https://YOUR-ENDPOINT/" s3 sync ./data s3://my-bucket/
Lots of small files are crawling¶
Per-object overhead dominates with tiny files. Increase concurrency, or archive small files into larger objects before transfer. A few large objects always move faster than thousands of tiny ones.
Could my region choice be the bottleneck?¶
Yes. High latency to a distant region lowers per-connection throughput. Pick the closest region; for new services, choose region by proximity to your users.
Is my own internet connection the limit?¶
Often, yes. You can't exceed your own local download capacity. Test from a well-connected machine, and compare against a speed test of your own link.
How should I benchmark properly?¶
Measure aggregate throughput across parallel transfers of representative data, not a
single cp. A single stream understates what your tier can do.
I've tuned everything and it's still slow¶
Email support@clearstreamer.com with your region, the command you ran, the concurrency you used, and the measured speed, and they'll check your dedicated node.