Kitz Forum

Broadband Related => Router Monitoring Software => Topic started by: adslmax on May 16, 2021, 01:22:10 PM

Title: Continuous Speed Test Software Tool
Post by: adslmax on May 16, 2021, 01:22:10 PM
I ran this continuous speed test software tool. Can be use in Windows, Linux and Android mobile. Download of 69Mbps and Upload of 11Mbps

Seem Windows not running full speed in real test only 69/11

Title: Re: Continuous Speed Test Software Tool
Post by: hacktrix2006 on May 16, 2021, 01:43:01 PM
Whats it called ? Might test it out myself.
Title: Re: Continuous Speed Test Software Tool
Post by: adslmax on May 16, 2021, 01:51:54 PM
Will get the link for you later when I get back home
Title: Re: Continuous Speed Test Software Tool
Post by: adslmax on May 16, 2021, 03:18:57 PM
Whats it called ? Might test it out myself.

There u go: http://startrinity.com/InternetQuality/ContinuousBandwidthTester.aspx
Title: Re: Continuous Speed Test Software Tool
Post by: banger on May 16, 2021, 08:30:10 PM
See the FAQ it downloads and uploads at the same time so you will get different speeds than other tests.
Title: Re: Continuous Speed Test Software Tool
Post by: Alex Atkin UK on May 17, 2021, 01:37:59 AM
I really don't see the logic in this unless you are trying to diagnose speed issues at certain times of the day.

You are effectively voluntarily making your connection unusable on a regular basis by downloading/uploading pointless data.
Title: Re: Continuous Speed Test Software Tool
Post by: banger on May 17, 2021, 02:22:56 AM
I really don't see the logic in this unless you are trying to diagnose speed issues at certain times of the day.

You are effectively voluntarily making your connection unusable on a regular basis by downloading/uploading pointless data.

You can adjust various things like download/upload with a pause. I am testing every 15 mins all day. Download /upload is low. Shame there are no graphs.
Title: Re: Continuous Speed Test Software Tool
Post by: tubaman on May 17, 2021, 07:48:53 AM
... I am testing every 15 mins all day...

As per @Alex Atkin UK, why would you want to do that unless trying to diagnose a specific issue. If too many people do this it'll just clog the backhaul system and nothing will work well.
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on May 17, 2021, 10:12:08 AM
I really don't understand the point of running this.

I'm aware of people who wrote scripts to achieve the same aim.

Looking for issues when there are none and degrading your own connection periodically seems crazy. Samknows runs at quiet times, this doesn't. It's unreliable due to actual, you know, use of the connection as something other than a speed tester, and disruptive to those actual uses.

Anyone using Zoom, etc, VoIP or gaming will notice degradation.

Just... Why? Who cares?

EDIT: Incidentally I note mention of P2P network in the servers list. This is running iPerf to a bunch of destinations and using your machine as one of those to measure performance to others.

Just in case that had been missed.
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on May 17, 2021, 12:40:33 PM
I've had the opportunity to have a look at this tool - for professional reasons actually, as customers may think it's a good idea to use it then raise issues with us.

http://startrinity.com/InternetQuality/ContinuousSpeedTesterFAQ.aspx

I'm now writing a knowledge base article to indicate to any less technical customers why using this as a measuring tool for, well, anything is a bad idea.

Thank you, Max, for pointing this out to us. I can save some of our support guys some time in the future.

https://iperf.fr between two endpoints you control is a far better idea. I can't say I envy anyone dealing with support calls from customers complaining about this tool reporting poor speeds when it ends up that they're testing between London and Singapore.
Title: Re: Continuous Speed Test Software Tool
Post by: Alex Atkin UK on May 17, 2021, 01:11:42 PM
I've fired up iperf3 on my VPS every now and again to check single-threaded speeds.  I can't see how testing against unknown servers is all that helpful.  Speedtest.net only get away with it due to the sheer volume of servers, even there its becoming less and less useful as people get Gigabit and the servers remain constrained on 10Gbit links.
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on May 18, 2021, 08:59:40 AM
The servers are just software on the Ookla side so can be easily upgraded as required. Most will have multiple 10G ports that can be aggregated.

A 4 lane PCI-E 4 card has more than enough capacity for 40G or 4 x 10G and storage is irrelevant for these purposes. 😊
Title: Re: Continuous Speed Test Software Tool
Post by: Alex Atkin UK on May 18, 2021, 10:40:00 AM
The servers are just software on the Ookla side so can be easily upgraded as required. Most will have multiple 10G ports that can be aggregated.

A 4 lane PCI-E 4 card has more than enough capacity for 40G or 4 x 10G and storage is irrelevant for these purposes. 😊

They can, but last I checked their rules were that to be valid as a 1Gigabit test server it only need a minimum of 10Gbit, and its not at all clear if/how they detect the usage of a server to ensure its hitting one with enough spare capacity.  I believe you yourself said in the past they can be very hit and miss on fast connections?

It just seems to me that residential speeds are increasing faster than server backhaul can be, so if people test too often then the chance of hitting a server with insufficient capacity is fairly high.
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on May 19, 2021, 10:46:46 AM
Yes, if people test obsessively then the whole thing comes tumbling down.

A fair few servers will end up closed off to preserve performance, reduce tests and maintain accuracy.

The ones I use are either not really widely known about or have a location that benefits me. Thankfully I only use them to test load balancing, check if there are issues and to annoy people.
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on June 03, 2021, 11:21:54 AM
Knowledge base article I wrote was used. Customer used this software and pointed at our software as the cause of the lower than expected performance.

A test to a known good cloud service produced the expected results. Happy days and thanks again for mentioning this.
Title: Re: Continuous Speed Test Software Tool
Post by: burakkucat on June 03, 2021, 04:57:45 PM
Knowledge base article I wrote was used.

Is your article publicly available, please? If yes, then I'm sure that quite a few of us would be interested in taking a look . . .
Title: Re: Continuous Speed Test Software Tool
Post by: adslmax on June 03, 2021, 05:42:37 PM
I have uninstalled this software from pc, laptop and my mobile no need it.  :thumbs:
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on June 04, 2021, 06:59:31 AM
Is your article publicly available, please? If yes, then I'm sure that quite a few of us would be interested in taking a look . . .

It is not: internal to assist our support folks, sorry. I'll copy/paste it next time I'm 'in' work.
Title: Re: Continuous Speed Test Software Tool
Post by: burakkucat on June 04, 2021, 04:20:32 PM
It is not: internal to assist our support folks, sorry. I'll copy/paste it next time I'm 'in' work.

Thank you. (Felines are always curious.)
Title: Re: Continuous Speed Test Software Tool
Post by: CarlT on June 07, 2021, 07:38:33 AM
Quote
Please be aware of 'Continuous Bandwidth Tester' - http://startrinity.com/InternetQuality/ContinuousBandwidthTester.aspx

The FAQ - http://startrinity.com/InternetQuality/ContinuousSpeedTesterFAQ.aspx - should tell you all you need to know. It appears to be shelled iPerf to random servers and even other users of the application on a peer to peer basis introducing a huge number of uncontrollable variables. Reports from residential users indicate it producing results lower than they would expect, which may have many causes.

Strongly recommend any time you guys see this, hopefully never, disregarding any results and pointing out to customers that it uses ‘multiple various servers around the world’ so there’s zero way of knowing where a bottleneck may be – testing between New York and Singapore has a few more variables entirely outside of the control of us, ISPs and indeed their ISPs than an iPerf across a fabric over the same connectivity provider.

This application cannot produce a representative view of the experience our customers will see in normal business operations. It cannot test overlay networks properly and I cannot see a legitimate use for it beyond for curiosity value - even then it would be very hard for it to satiate that curiosity with accuracy.

For performance across a fabric the in-house iPerf test is the way to go. For performance to the Internet directly the gold standard remains iPerf to a server under control and monitoring of the user. For performance testing of backhauled traffic iPerf across a fabric incorporating a cloud platform. Alongside these reports from the regular telemetry suites and UX should work.

For larger providers with extensive residential arms likely to have Netflix caches and for simplicity https://fast.com can be useful as a very basic confirmation of Internet bandwidth but, as per, keep the test relevant to what is being tested and eliminate as many variables as possible.

Lastly, running a tool like this may cause issues where QoS is misconfigured. Intentionally saturating with pointless traffic that may have been configured to be high priority as it could be considered telemetry will both degrade lower priority traffic and degrade the quality of the telemetry being produced. Even if not run constantly when it is run if set too close to circuit saturation levels it will generate microbursts that may cause issues with local QoS and buffering on directly connected upstream devices.
Title: Re: Continuous Speed Test Software Tool
Post by: burakkucat on June 07, 2021, 03:35:21 PM
Thank you for posting a copy of your article, Carl. Hopefully it will be a sufficient warning for any of our members who might, in the future, (re-)discover that flawed "tester".