I found a better speed test for satellite internet

[grohgreg]

tell me the test id please.

Last one I performed was #1218



//greg//

[grohgreg]

Hughes is intentionally dropping packets to achieve the service level you're seeing. Some may call it throttling, comcast users know it well as torrent busting and similar nasty names.

Maybe it looks like that to you, but that's not really what's happening. Remember, this is burst transmission shared bandwidth - so a highly interrupted serial bitstream is the rule rather than the exception. Whereas it may appear as dropped packets to you, it's actually gaps in the burst transmissions. Think of it as a buncha folks lined up outside a buffet restaurant at dinner time. They get let in to feed - one at a time - then are seated at a common table to eat. But don't even think about getting up to pee, cuz the next guy in line will jump right in your seat. You have to go to the back of the line and work your way forward again. This can happen even while the distant server is waiting for an ACK. If someone else gets your slot, it looks like a NACK to the distant server. To mix the metaphors even more, the longer the queue - the greater the (time) gap between individual customer bursts.

With satcom, the problem is compounded by varying transmit signal strengths. The guy with the strongest carrier can actually have the effect of bumping weaker ones out of the queue. This also goes back to where I said the simple throughput over time method is all but useless for us. It often counts the initial lag (latency) - and the dead spots between bursts - against the time component.

And that's just the gateway server. That's where the packet headers are read, then prioritized/routed according to protocol detected. Most go first to a DNS server, HTTP also goes through a HPEP server (accelerator), everything else (FTP, UDP, etc) gets pushed out into the internet. The stuff that the consumer grade providers don't like (P2P, VoIP, streaming, etc) gets treated like the proverbial stepchild, receiving lower processing priority. I guess the corporate logic is that if they render the performance of this kinda stuff so poor, some "bandwidth hogs" may just give up trying to use it over satellite.

On the flip side, this is all technically reversible. For those willing to pony up the bucks, business grade service (as opposed to consumer grade) can actually be customized to prioritize by specific protocols. One can specify VPN and/or VoIP acceleration (as opposed to suppression), even HTTPS acceleration can be made available. Keep in mind I'm speaking of outfits like Hughes and iDirect. I don't think what Wildblue VARs sell as business/enterprise service is that technically advanced yet

//greg//

[Bandwidth]

Last one I performed was #1218

//greg//

We got it and found your others based on the ip. #1218 wasn't all that bad if I remember, will look again. Thanks

[Bandwidth]

Maybe it looks like that to you, but that's not really what's happening. Remember, this is burst transmission shared bandwidth - so a highly interrupted serial bitstream is the rule rather than the exception. Whereas it may appear as dropped packets to you, it's actually gaps in the burst transmissions. Think of it as a buncha folks lined up outside a buffet restaurant at dinner time. They get let in to feed - one at a time - then are seated at a common table to eat. But don't even think about getting up to pee, cuz the next guy in line will jump right in your seat. You have to go to the back of the line and work your way forward again. This can happen even while the distant server is waiting for an ACK. If someone else gets your slot, it looks like a NACK to the distant server. To mix the metaphors even more, the longer the queue - the greater the (time) gap between individual customer bursts.

With satcom, the problem is compounded by varying transmit signal strengths. The guy with the strongest carrier can actually have the effect of bumping weaker ones out of the queue. This also goes back to where I said the simple throughput over time method is all but useless for us. It often counts the initial lag (latency) - and the dead spots between bursts - against the time component.

And that's just the gateway server. That's where the packet headers are read, then prioritized/routed according to protocol detected. Most go first to a DNS server, HTTP also goes through a HPEP server (accelerator), everything else (FTP, UDP, etc) gets pushed out into the internet. The stuff that the consumer grade providers don't like (P2P, VoIP, streaming, etc) gets treated like the proverbial stepchild, receiving lower processing priority. I guess the corporate logic is that if they render the performance of this kinda stuff so poor, some "bandwidth hogs" may just give up trying to use it over satellite.

On the flip side, this is all technically reversible. For those willing to pony up the bucks, business grade service (as opposed to consumer grade) can actually be customized to prioritize by specific protocols. One can specify VPN and/or VoIP acceleration (as opposed to suppression), even HTTPS acceleration can be made available. Keep in mind I'm speaking of outfits like Hughes and iDirect. I don't think what Wildblue VARs sell as business/enterprise service is that technically advanced yet

//greg//

Pull the detailed analysis of the TCP speed test and you will see exactly how they throttle or in this case manage traffic. Give it a shot and let me know if it doesn't demonstrate exactly what we're talking about. Feel free to keep your results links for comparison, they stay active at least 90 days. I'm working on a way you can pull results in a number of formats including excel feedback from the user is needed about the hostname and masks.

I've asked the question on our site and I'll ask it here. If we give full access to the reporting engine does anyone object to ip's being shown? Honestly this should be a no-brainer because no personal information is obtained and everyone knows your ip these days anyway. Depending on the answers posted I may seek permission to open up the reporting feature to everyone.

[grohgreg]

We got it and found your others based on the ip.

Fine. Now perhaps you'll realize that I'm not in this for the fun of it. But it should also go to demonstrate the monkey wrench that a NAT server can throw into the mix.

If we give full access to the reporting engine does anyone object to ip's being shown?

No sweat here. All you're going to "expose" is the address of a Hughes NAT server

//greg//

[tobifelinis]

I don't care mines changed 4 times today already.

[Bandwidth]

Alright here is the url to pull your own stats. It's different and you will need to play a little to customize it for your needs. You can pull results by any number of data types and then of course import them into excel and all the usual things.


MyConnection Server Report

Knowing your current IP will be helpful. Because satellite users appear to change IP's often (we just noticed this) we're going to enable session id's for just the satellite tests. This will allow you to enter a unique id just for your tests. So long as you remain consistent with your session id all your tests will be available.

[grohgreg]

Because satellite users appear to change IP's often (we just noticed this) ....

So then you don't actually know what a NAT server is ???

//greg//

[Bandwidth]

So then you don't actually know what a NAT server is ???

//greg//

Greg...you have a way with words that is really startin to piss me off. Stop looking for fault in everyone. You have a communication style that at best is venomous and you really need to check it. I have no problems answering your questions but I WILL NOT tolerate this "ooooo I got you now" attitude and unlike porker I will take personal pleasure in appropriately taking you to the mat. So...let's go back to that positive and constructive conversation now...shall we....

Alright then, not only am I familiar with NAT but have implemented several similar systems current and past. I hold several long standing certifications and almost 22 years in the network engineering field. Remember I also had Direcway at one time and this is probably one of the only GREAT features about it. For some its a pain, I loved it.

The surprise was that connections from the same origination point within just a couple seconds from each other resulted in different ips (I don't care if it's your actual, we both know it's not). I expected things to remain at least somewhat constant for at least 30 seconds. And they didn't.

[tobifelinis]

Did youchange something on the Route Performance test? Yesterday it was showing the NOC firewall and today it does not but is showing something that i make as the proxy server.