Participant

Why Different Speeds At Different Test Sites.

I go to Spectrum Speed test site and do a test. I see 175 down and 6 up. Suposed to have 200 down and 10 up so, go to Ookla and run a test using the same site IP as Spectrum and see 231 up and 12 down. Spectrum testing is also provided by Ookla; I go then to 3 other test providers and see average of 220 up and 11 down with all above 200/10. Guess I avoided a service call but why the difference especialy with same provider and IP?

 

John

5 REPLIES 5
Proven Sharer

Re: Why Different Speeds At Different Test Sites.


@n4zxwrote:

I go to Spectrum Speed test site and do a test. I see 175 down and 6 up. Suposed to have 200 down and 10 up so, go to Ookla and run a test using the same site IP as Spectrum and see 231 up and 12 down. Spectrum testing is also provided by Ookla; I go then to 3 other test providers and see average of 220 up and 11 down with all above 200/10. Guess I avoided a service call but why the difference especialy with same provider and IP?

 

John


Spectrum's testing uses Ookla software installed on Spectrum servers at several locations across the country.  Those sites are very near to the IXPs who take traffic from Spectrum and send it to the internet.  So when you use the Spectrum Speedtest, you are measuring speed of Spectrum's private transport network between your computing device and the selected IXP's portal to the internet.  This will always be faster than readings to server sites located further out in the internet. 

Sharer

Re: Why Different Speeds At Different Test Sites.

Some routes are simply cleaner/faster than others. Go back to the speedtest site, and search for Montreal and test each provider on the list. Can be a real eye opener how much things can vary within just the same city.
Proven Sharer

Re: Why Different Speeds At Different Test Sites.


@RAIST5150wrote:
Some routes are simply cleaner/faster than others. Go back to the speedtest site, and search for Montreal and test each provider on the list. Can be a real eye opener how much things can vary within just the same city.

But again, Montreal is outside Spectrum's private network, and they are not responsible for any equipment performance or resulting loss of speed beyond the edge of their system.  It's not like they use dynamic path routing, redundancy, and/or load sharing to always deliver the fastest data speed between any two individual IP addresses.  All internet traffic  moves through just a handful of IXP gateways across the USA which are preselected primarily by distance to the subscriber's local operating franchise.  When you chat with your neighbor who uses a different ISP, the path can be hundreds or even thousand of miles.  

Sharer

Re: Why Different Speeds At Different Test Sites.

The point I was trying to get to is more about how frequently routing can differ getting to what may appear to be the same general location.

The OP was actually questioning why the results were the opposite of what you mentioned--got better test results when stepping outside of Spectrum's IPX. Note the results improved when they went to the public speed test site and tested against locations selected there--compared to testing to through Spectrum's dedicated application. BGP may have wound up offering a cleaner route for the public test versus the internal one.

Just because you think you may be testing against the same facility in the same town, it doesn't necessarily mean you are taking the exact same route with each test--or even hitting the same IP/server at the endpoint for that matter. Even if it involves the same peering partner(s) along the way, it can still be different devices in play that are coming under varying levels of utilization. Packets could get held in queues longer, and so and so and so forth.

That is why I mentioned testing specifically to Montreal. I was one of a group of players that worked at length for months at a time with our own Tier3 teams as well as the ISP at the other end addressing route stability issues for FFXIV when it was being hosted up there. As we regularly gauged the landscape around Montreal and performance across mutliple peering partners, it was not uncommon to see large variances in latency from one carrier to the next, not to mention a similar impact to throughput as well--even though we were targeting the same locations. It was just a convenient example for demonstrating things are not always as simple as they seem when it comes to internet routing.

Without taking more specific approaches in testing, you are rarely able to reduce it to such a simple apples to apples comparison... especially not with just a few simple tests between what on the surface appears to be the same locations. There is a lot going on under the hood that most of us are completely unaware of.
Participant

Re: Why Different Speeds At Different Test Sites.

Thanks to all for the information. My head hurts for processing itSmiley Happy My original point is that while differences in speed can be explained as noted by y'all the ones I see 150- 175 up when using the recommended test by Spectrum are greatly different than what is seen by other sites even going to the same IP. 50 to 75 Mbp ( Spectrum says 175 and Ookla says 231 for example) is a huge difference even accounting for the noted reasons and led to unnecessary service calls and truck roles. Another thin Spectrum can be as low as 5 and 50 but other sites are consistent in re-measuring. Thanks again