cancel
Showing results for 
Search instead for 
Did you mean: 

Testing latency: Dallas compared to Amazon

maratsade
Distinguished Professor IV

Re: Testing latency: Dallas compared to Amazon

Everyone is seeing this. The difference is, if you have cable, you have  higher speeds, so you notice it less.  @MarkJFine has shown the problem with CL many times. 

And on the home subscriber side of CL, user complaints are legion. 

MarkJFine
Associate Professor

Re: Testing latency: Dallas compared to Amazon

As an example, my web server has started timing out when backing up to Google Drive because of complete gaps in service of over 60 seconds - all over a CL-only path.


* Disclaimer: I am a HughesNet customer and not a HughesNet employee. All of my comments are my own and do not necessarily represent HughesNet in any way.
lighthope1
Senior

Re: Testing latency: Dallas compared to Amazon


@maratsade wrote:

Everyone is seeing this. The difference is, if you have cable, you have  higher speeds, so you notice it less. 


It's not a speed problem.  It's a latency problem.   My speeds are 43mbs.  And I am still getting multi-second latency.

 

Are you saying that CenturyLink customers are getting multi-second latency spikes all the time?  (I haven't heard any of those complains, but I could be looking in the wrong place.)

 

maratsade
Distinguished Professor IV

Re: Testing latency: Dallas compared to Amazon

It's not just latency; it's more complex than that and it involves issues with the backbone,  network load, task prioritisation, congestion, and probably other variables. All those together may be interpreted by users as latency issues or speed issues, but it's more than that.  It's also very noticeable if you're on satellite, but it's affecting everyone. 

lighthope1
Senior

Re: Testing latency: Dallas compared to Amazon


@maratsade wrote:

It's not just latency; it's more complex than that and it involves issues with the backbone,  network load, task prioritisation, congestion, and probably other variables. All those together may be interpreted by users as latency issues or speed issues, but it's more than that.  It's also very noticeable if you're on satellite, but it's affecting everyone. 


WHY is it more noticable on satellite?

 

Consider this:  There are five hops from me to the destination.

 

1  2  3  4  5

 

Now, at hop 2 is CenturyLink.  There is a 70ms delay due to traffic shaping.  So that adds 70ms to the delay.

 

Hop 3 is another CenturyLink.  There is a 100ms delay due to a problem with their server.  Now we're up to an additional 170ms delay.

 

Why is a satellite user seeing a 5000ms delay when a terrestrial user is only seeing a 200ms delay?

 

Unless for some reason hop 2 and 4 is communicating back to the user.  Thus you're having to resend the data two, three, four times.  That would certainly account for it.

 

That's not a delay, that is a packet loss.  (I may be using the wrong term here.)

 

If CL is losing data and requiring us to resend it, that is a major issue.  Nothing to do with latency.

maratsade
Distinguished Professor IV

Re: Testing latency: Dallas compared to Amazon

"WHY is it more noticable on satellite?"

Because satellite is slower and has more latency, so issues stand out a lot more, kind of like a guy with a big mustache stands out more in a small room with 15 people in it than in a stadium with 20,000.  

"Why is a satellite user seeing a 5000ms delay when a terrestrial user is only seeing a 200ms delay?"

My guess would be that it has something to do with the pathways between CL and individual ISP servers, so there are different routes your request would follow, depending on the ISP.  Some pathways may be slightly cleaner, less congested, etc.  EDIT: I imagine that the fact that our signals travel to and from space has something to do with the issue too. 

CL does have major issues.