I found a better speed test for satellite internet

**grohgreg** · 05-11-2009 08:32 AM

Originally Posted by Scott Greczkowski

BTW if anyone wants to test their speed, we have a speedtest here on our servers.

Speedtest.net Mini Bandwidth Speed Test
Its actually been there for about 2 years.

I find it interesting Scott, that this is same test as one of those under scrutiny in this topic. But I assume you just rent the right to use somebody else's product. Not surprisingly, the mini-test demonstrates the identical issue with upload speed results. Specifically, results are demonstrably inflated when simultaneously compared to the bitstream at point of origin. For example; I might be sending a measured 170k, the test site reports back 250k. The mini-server is seeing a proprietary bitstream on the satellite side of the modem - which is compressed and FEC encoded and already suffering from at least 500ms of lag. On the other hand, I can simultaneously sample the raw bitstream on the WAN - between the modem and the PC. And as you know, the raw bitstream is the one that defines true operating speed. Inroute overhead isn't added until processed by the subscriber modem. Outroute analysis is similar, but usually employs far less FEC - so needs a separate speed test algorithm of its own.

Besides, the upload test file is grossly undersized. The (minimum) half second lag alone makes a simple throughput over time calculation wildly inappropriate. But that's the easy part. The real stumbling block is how to accurately factor in the various and proprietary transmission methods. You likely already understand burst transmission. But for those that don't, the gateway server must be given sufficient time to set a throttle appropriate to the rate plan specific to that bitstream. As such, the initial transmission will be a burst that is considerably in excess of that permitted by the rate plan. It however is passed through to the test server. But in this case, the upload test file itself is so small - and the lag is so great - that the test is over before the properly throttled bitstream gets properly factored into the test results. It's long been accepted that 300k is the smallest file size recommended for typical consumer grade satellite inroute testing. But the real cherry on top is that both TX and RX data rate and coding are adaptive. At least that's true for my provider. Specifically, it's possible (for my modem) to shift data and/or FEC rates in mid-test; a factor that simply can NOT be taken into consideration by a 3rd party test server.

That said, the fact that each satellite provider employs a proprietary transmission method - works against the speed test provider. It's just not practical to tailor individual speed tests on a provider basis. The guy that thinks he's going to come up with a universally accurate speed test for ALL satellite connections - is tilting at windmills. Which is yet another reason to stick with diagnostic tools specific to the satellite provider.

Anyway. I also find it interesting that bwporker (which I assume stands for Bandwidth Hog) and Bandwidth are both Floridians. Given the similarity of their screen names and coincidental proximity, I'm curious if they also post from the same IP address. Have a good time in Denver by the way !!

//greg//

**bwporker** · 05-11-2009 11:08 AM

Originally Posted by grohgreg

Anyway. I also find it interesting that bwporker (which I assume stands for Bandwidth Hog) and Bandwidth are both Floridians. Given the similarity of their screen names and coincidental proximity, I'm curious if they also post from the same IP address. Have a good time in Denver by the way !!

//greg//

Greg,

Denver is for wooses. Nothing but yuppified liberals in Denver. I prefer Colorado in it's natural state, untouched and I'm not telling you where because to be honest you have that annoying stalker trait and it's none of your business.

**grohgreg** · 05-11-2009 11:50 AM

Folks that actually read before they respond would have noticed that my last post was to Scott (the owner of this website). This is not now, nor has it ever been - all about you !!

//greg//

**bwporker** · 05-11-2009 12:00 PM

Originally Posted by grohgreg

Folks that actually read before they respond would have noticed that my last post was to Scott (the owner of this website). This is not now, nor has it ever been - all about you !!

//greg//

Now I really don't want to give you any personal information. Not only are you a forum stalker but apparently you are bi-polar as well. From the start you've been attacking me and your previous posts verify it. I don't know and I don't care why you are behaving this way but I have no intentions of leaving this site because "YOU" don't like me. I made a post that I thought would help people and instead you have managed to dominate the thread with your personal attacks....go away kiddie and let people who really want to learn and make things better communicate with each other without fear of reprisal from the almighty Greg.

**tobifelinis** · 05-11-2009 02:04 PM

Well I ain't busy and just for the heck of it I just ran three speed test on IP geeks and Hughes.

I never had much for speed test and still don't. My plan says I get 1000 down 128up.

So Ip Geeks:
1070/105
1016/105
1083/117

Hughes server

1016/151
1008/160
990/140

Don't think it proves anything.

I think the VOIP test might be valid can't really tell but it might be a good place to send the Hughes folks who don't understand why VOIP don't work on Hughes. Dazzel em with charts.

**tobifelinis** · 05-11-2009 02:22 PM

Incase anyone cares here is the path test clearly stopping at the NOC.

**grohgreg** · 05-11-2009 03:01 PM

Originally Posted by bwporker

Now I really don't want to give you any personal information. Not only are you a forum stalker but apparently you are bi-polar as well. From the start you've been attacking me .....

Interesting take - but it's still all about you, isn't it?

Let me tell you how things work in the real world amigo. You popped in here as a freshman poster - almost certainly without any research at all - and started promoting a website that I'm still not sure you understand to this day. I gave you the benefit of the doubt, and provided personal observations to the contrary. No brag, just fact.

But somehow you became offended because your beneficent (but uninformed) gesture wasn't appreciated. If you go back over the sequence of events here, it's pretty clear who's the actual name-caller. And the fact that "Bandwidth" feels it's ME who owes you an apology, still makes me think you're posting under two member names.

Who asked you for personal information anyway? The bigger question is - who cares? And since you're the freshman here, how on earth could I be the stalker? You came into MY house !!!

You've got some serious personal issues there mate, and this is NOT the place to air them.

//greg//

**grohgreg** · 05-11-2009 03:03 PM

Originally Posted by tobifelinis

Incase anyone cares here is the path test clearly stopping at the NOC.

Good example Tobi. It simply reinforces what I posted earlier about these terrestrial speed test sites not typically coping with the NAT server concept

//greg//

**Bandwidth** · 05-11-2009 03:49 PM

Originally Posted by grohgreg

I find it interesting Scott, that this is same test as one of those under scrutiny in this topic. But I assume you just rent the right to use somebody else's product. Not surprisingly, the mini-test demonstrates the identical issue with upload speed results. Specifically, results are demonstrably inflated when simultaneously compared to the bitstream at point of origin. For example; I might be sending a measured 170k, the test site reports back 250k. The mini-server is seeing a proprietary bitstream on the satellite side of the modem - which is compressed and FEC encoded and already suffering from at least 500ms of lag. On the other hand, I can simultaneously sample the raw bitstream on the WAN - between the modem and the PC. And as you know, the raw bitstream is the one that defines true operating speed. Inroute overhead isn't added until processed by the subscriber modem. Outroute analysis is similar, but usually employs far less FEC - so needs a separate speed test algorithm of its own.

Besides, the upload test file is grossly undersized. The (minimum) half second lag alone makes a simple throughput over time calculation wildly inappropriate. But that's the easy part. The real stumbling block is how to accurately factor in the various and proprietary transmission methods. You likely already understand burst transmission. But for those that don't, the gateway server must be given sufficient time to set a throttle appropriate to the rate plan specific to that bitstream. As such, the initial transmission will be a burst that is considerably in excess of that permitted by the rate plan. It however is passed through to the test server. But in this case, the upload test file itself is so small - and the lag is so great - that the test is over before the properly throttled bitstream gets properly factored into the test results. It's long been accepted that 300k is the smallest file size recommended for typical consumer grade satellite inroute testing. But the real cherry on top is that both TX and RX data rate and coding are adaptive. At least that's true for my provider. Specifically, it's possible (for my modem) to shift data and/or FEC rates in mid-test; a factor that simply can NOT be taken into consideration by a 3rd party test server.
//greg//

Everything you've stated above is correct. Let's answer most of the concerns you've raised and see if you're happy with the answers. The speedtest mini applet that Scott uses for his general test can be obtained from the Ookla website and is intended for webmasters. There is no charge and it must be installed on a decent server and broadband pipe (meaning no bandwidth limitations or issues). Scott's server peformed well in several tests that I performed and I told him so. We also offer the applet on several servers throughout the country but we discourage use of them for diagnostic purposes. They are intended to give you instant feedback on your approximate bandwidth capability at that moment.

The speedtest mini applet was not intended to benchmark satellite or hybrid broadband connections and honestly shouldn't be relied upon too much. For many broadband users, especially those using any form of acceleration technology the tests would also not be appropriate.

With regards to our site, unfortunately bwporker posted the wrong test, results and link to information. Because the test is labled as SPEED people assume this is what the test does. This is a common mistake and we've tried to educate users since making the tests available to the public. Similar limited versions of the tests on other sites also are being misused. The webmasters at each of those sites are no doubt confused about the purpose of that test. Admittedly I think we could have done a better job of explaining the difference and proper use and we'll work on that in the coming days.

The proper test in our case is the ISP Capacity Test located under Diagnostic Tools > ISP Capacity Test. As it's offered TODAY this test is not optimized for satellite however this is about to change much in part due to this thread. Until then we're confident the test will provide consistent results that certainly can be relied upon. Your concern about rate, overhead, burst and test duration are all addressed in this test. Burst and cache detection are inluded and noted on the results where applicable.

Hopefully we can put this behind us and move forward. We encourage anyone to provide feedback and will always respond to comments regarding any of the tests. Thanks again for the opportunity to clear this up.

**Bandwidth** · 05-11-2009 04:01 PM

Originally Posted by grohgreg

Interesting take - but it's still all about you, isn't it?

Let me tell you how things work in the real world amigo. You popped in here as a freshman poster - almost certainly without any research at all - and started promoting a website that I'm still not sure you understand to this day. I gave you the benefit of the doubt, and provided personal observations to the contrary. No brag, just fact.

But somehow you became offended because your beneficent (but uninformed) gesture wasn't appreciated. If you go back over the sequence of events here, it's pretty clear who's the actual name-caller. And the fact that "Bandwidth" feels it's ME who owes you an apology, still makes me think you're posting under two member names.

Who asked you for personal information anyway? The bigger question is - who cares? And since you're the freshman here, how on earth could I be the stalker? You came into MY house !!!

You've got some serious personal issues there mate, and this is NOT the place to air them.

//greg//

This is not YOUR house, this is Scott's house. Either leave the guy alone or I will take the issue up with Scott myself admin to admin if bwporker doesn't. Got it Greg? or do I need to spell it out for you? Now enough of the flaming!