Network upstream/downstream speeds

August 27, 2014 5.7k views

How can I measure what's the maximum/minimum download and upload speed on the droplets? I tried with speedtest-cli, but the reported speeds vary a lot. Is this a variable number that varies depending on uncontrollable conditions?

My intention is to deploy a VoIP application that will be consuming a regular amount of bandwidth and I am trying to determine how many simultaneous calls can run on a single droplet.


2 Answers

I also noted, that I have better upload speed at NYC1 datacenter, compared to SF1. Ping times are also better for NYC1, than SF1.

I am in Mexico, and I thought it would be better to use a region closer to where the service is gonna be deployed, so this is making little sense to me now. Am I missing something?

Also, it may be important to notice that my SF1 is a 5$/mo droplet, and my NYC1 is a 20$/mo droplet. Does this affect the overall network performance?

  • Network speed should not vary based on the price of the plan. From the FAQ:

    All DigitalOcean physical nodes are connected via gigabit ethernet to switches, which are uplinked to our aggregators and core routers. All core routers are uplinked via 10-gigE uplinks to providers.

    The reason for the difference in ping times may have to do with how the requests are being routed by network providers. If you're experiencing a lot of latency from SFO, please open a support ticket that includes a traceroute so the team can investigate.

Thank you for your answer Andrew.

Running traceroute showed 20 hops for both regions, but as I mentioned, ping times are different in favor of NYC, which is way farther than SF measuring from Mexico City.

  • To SF: ~100ms
  • To NYC: ~90ms


Have another answer? Share your knowledge.