I'm conducting Ookla speedtests from AWS in us-east-1 to multiple Ookla speedtest servers in Singapore and I'm finding that upload speeds are consistently much lower than download speeds, by a factor of 5-10x. (e.g. download will show as 500 Mbps and upload will show as 50 Mbps)
My test setup: - AWS us-east-1 region - Windows Server 2019 - Ookla CLI for Windows (i.e. not the browser version of Ookla) - t2.xlarge instance (supports up to 1 Gbps)
I understand that I should see some degradation in speed the further away my client is from the Ookla server but what I don't understand is why the upload speed is so much lower than the download.
This isn't a home network where there is asymmetry between upload and download. I have tested with multiple nearby Ookla servers in us-east in Ashburn, VA and then I get consistent 980 Mbps up and down speeds so I'm confident I should be able to get symmetrical speeds when using further away Ookla servers.
But as soon as I test with Ookla servers far away from my client in us-east-1, the upload speed drops drastically from the download speed. I have tested this with multiple Ookla servers in each test region so I'm confident this isn't an issue with a single Ookla server.
I have also setup an iPerf test from us-east-1 to a Digital Ocean server in Singapore and have been able to achieve 1 Gbps upload and download:
iperf3 -c <public IP of server> -u -b 1G -t 30
At this point, it seems like there's an issue with how Ookla does their upload tests because I was able to get fully symmetrical tests with iPerf but not with Ookla.
Am I missing something here? Or are Ookla speedtests not reliable?
No comments:
Post a Comment