All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Why are upload/iperf tests so rare from providers?
Hello, LET Community,
So here's something I've been wondering about, and I figured this would be the best place to get some thoughts. I was shopping for a new server for a backup project, and the lack of upload testing has become a real sticking point for me.
When we're shopping for a new VPS or dedicated server, why do the vast majority of hosting providers only offer download and ping tests?
It seems standard practice to provide a test file to check download speeds and a test IP for ping/latency checks. However, it's surprisingly rare to find a provider that offers a way to test upload speeds before buying. Going a step further, tools like iperf, which provide a much more thorough measure of network throughput, are almost never offered.
The main exception I've consistently seen is OVH. They provide a comprehensive set of network testing tools, including public iperf3 servers at their various datacenters. This allows potential customers to get a much clearer picture of both download and upload capabilities before committing to a purchase.
This leads me to a few questions for the community and providers here:
Is it a matter of demand? Is the primary focus on download speed because that's what most customers use for things like web hosting? For many of us, our primary interaction is uploading content to the server in the first place. While download speed is obviously important, many of us also rely on solid upload speeds for backups, streaming, running applications, or even seeding... well, Linux ISOs.
Are there technical or cost concerns? Would offering a public iperf3 server put too much strain on the network or host node? Is there a risk of abuse that providers are trying to avoid?
Is it about managing expectations? On a shared node, upload speeds can be more variable than download speeds. Perhaps by not providing an easy way to test, providers avoid complaints if the speeds don't consistently hit a certain number?
I get that running a public iperf server 24/7 might have its downsides. But what about offering it on-demand? Perhaps providers could offer a temporary test via a support ticket for serious potential customers?
Finally, for other users, how do you solve this pre-sales puzzle? Since we can't run scripts like YABS before buying, what's your strategy? Do you rely heavily on recent community reviews? Do you specifically look for providers with a solid money-back guarantee, or even better, a true trial period? I recently saw Layer7.net offers a 48-hour trial, which seems like an excellent solution to this problem. Or have you ever had luck asking a provider's pre-sales support directly for a temporary test or a screenshot of an iperf result?
Thanks
Comments
In some of our locations where we have a VM with a public speedtest/iperf3 server (https://lg-mci.advinservers.com), we notice network usage between 50-100MB/s on average. We don't have a problem with it because we have excess bandwidth, but I can see why some providers might not want to have it.
That could be $25-500+ per month in bandwidth depending on the location.
Adding another thought to this: it's not just about upload vs. download, but also advertised port speed vs. real-world throughput.
We see "10Gbps Port" advertised everywhere now. But what good is that if the provider's peering is poor or their transit links are congested?
Many times, I've seen servers with high-spec ports struggle to push even 200Mbps to certain locations. (I know it’s unrealistic to expect a perfect connection to every corner of the world.) An iperf test would reveal this instantly, cutting through the marketing hype to show the real, end-to-end performance.
So maybe that's another reason providers avoid it? It would expose that a big port number doesn't always equal a fast connection in practice.
the reason is simple
most looking glass software especially open source one doesn't have upload from user functions.
you think anyone would just go all the way out to implement those and spending time to figure out how to deal with abuse potential? lol.
Honestly? I think most don't care enough to check. The vast majority of customers, even those running production environments on their servers, aren't stressing the ports for a test unless they specifically need high throughput for some niche or specific use case.
I'd imagine that once you get into offering 10Gbps+ dedicated then it becomes more likely that users would want or need to test things, but high bandwidth / fat port hosting is still relatively niche and not a need of the average consumer.