Geekzone: technology news, blogs, forums
Guest
Welcome Guest.
You haven't logged in yet. If you don't have an account you can register now.


Filter this topic showing only the reply marked as answer View this topic in a long page with up to 500 replies per page Create new topic
1 | 2 | 3
5187 posts

Uber Geek
+1 received by user: 1681


  Reply # 775113 5-Mar-2013 16:56
Send private message

kentnl: [
Its nice that TCL have a 100M plan, but I've yet to meet anyone on that plan who are actually getting what they're paying for. Its great to have 100M for local connections, ie: to Snaps test server .... but having it drop to 10% for ALL international traffic just makes that extra payment a joke.


If you are not getting what you are paying for, then raise it with them and complain. However, as Johnr noted, unless you are on a business plan with an SLA, then you are getting what you pay for. There is no guaranteed speed for a single international connection - this is affected by many things out of TC's control, including the server at the other end you are testing from.

Try multi threading your downloads and see if you can get increased throughput.

19282 posts

Uber Geek
+1 received by user: 2600
Inactive user


  Reply # 775117 5-Mar-2013 17:01
Send private message

Or drop to a lower speed plan

 
 
 
 


23 posts

Geek
+1 received by user: 4


  Reply # 775122 5-Mar-2013 17:08
Send private message

RunningMan:

If you are not getting what you are paying for, then raise it with them and complain.


Yeah, in progress queries trying to find better routes


However, as Johnr noted, unless you are on a business plan with an SLA, then you are getting what you pay for.


If you have a plan that says "100M on the box" and says "You won't get this speed all the time", sure, but theres the implicit notion you'll get that speed occasionally, not never.

ie: When you sell a plan that has 10M, you don't expect that this means all customers will only get dial up speeds all the time, because thats the degree of difference we're seeing here.

Hell, if you know your average consumer isn't going to get more than 20M/s downstream, why have plans that offer more, that cost more, instead of just having 20M plans? At least that way nobody is going to complain they're not getting what they're paying for.


There is no guaranteed speed for a single international connection - this is affected by many things out of TC's control, including the server at the other end you are testing from.


Sure, if the server at the other end is limiting it, that will be a problem, but the speed test servers I'm connecting are not limiting the speed at their end, and they have regular users who often break the 100M margin. Hell, I know that there have even been some 100M connections from NZ.


Try multi threading your downloads and see if you can get increased throughput.


Multi-segmented downloads seem to get better throughputs some of the time, but usually, if multi-segmented transfers get you better speeds than single-threaded, that suggests something is limiting throughput arbitrarily in-between.




23 posts

Geek
+1 received by user: 4


  Reply # 775124 5-Mar-2013 17:13
Send private message

johnr: Or drop to a lower speed plan


Unfortunately, the lower speed plans of TelstraClear also have lower package sizes, so you can't opt for a 180G download package and a lower speed plan.



Perhaps you could argue that this essentially means we're not paying for the speed, but for the download package size, and I'd have to try divining how much we're paying for the "speed" part of the plan.

Perhaps its just a bad choice in marketing on TelstraClears behalf, but the way its worded and marketed is very centric on "This plan is much faster than the other plans" which in truth, it isn't.



3759 posts

Uber Geek
+1 received by user: 2275

Trusted
Spark NZ

  Reply # 775125 5-Mar-2013 17:20
Send private message

kentnl:[snip]

If you have a plan that says "100M on the box" and says "You won't get this speed all the time", sure, but theres the implicit notion you'll get that speed occasionally, not never.


kentnl:
Last Result:
Download Speed: 98823 kbps (12352.9 KB/sec transfer rate)
Upload Speed: 9472 kbps (1184 KB/sec transfer rate)
Latency: 21 ms
Tue 05 Mar 2013 16:39:56 NZDT


OK, 98823 kbps isn't 100m, I grant you, but it's close. Expecting any connection to actually reach it's access speed to ANYWHERE beyond the local segment (or perhaps the national extent) of a providers network is unrealistic.

You can desire all you like for single threaded 100mbps downloads from Germany or the US, but I'm afraid reality is different.

Cheers - N


23 posts

Geek
+1 received by user: 4


  Reply # 775139 5-Mar-2013 18:12
Send private message



OK, 98823 kbps isn't 100m, I grant you, but it's close. Expecting any connection to actually reach it's access speed to ANYWHERE beyond the local segment (or perhaps the national extent) of a providers network is unrealistic.

You can desire all you like for single threaded 100mbps downloads from Germany or the US, but I'm afraid reality is different.

Cheers - N



The Expectation is that yes, there is traffic limitations due to having to compete for bandwidth at peak times, but at non-peak times, I'd rather have a better technical explanation of why this speed can't be achieved.

Because I'm certain, that If I were to lease 10 connections in the same location, each with the promise of 100M, but each with only 10M delivered,  I could easily amass 100M aggregate throughput during non-peak time.

And the point remains; why do I get poorer speeds on a cable connection, with no intrinsic speed limitation, than my ADSL friends, to the same destinations, when ADSL has technical limitations on the medium, and I do not?

784 posts

Ultimate Geek
+1 received by user: 103
Inactive user


  Reply # 775160 5-Mar-2013 18:30
Send private message

I'm also on TCL Warp and have also noticed international speeds generally getting slower. Whilst i do get what I am contracted for in terms of national traffic, ISP's must know that most of us require international traffic.

I believe that it is this that ISP's play with regularly to adjust their costs. 

5187 posts

Uber Geek
+1 received by user: 1681


  Reply # 775192 5-Mar-2013 19:22
Send private message

kentnl:
The Expectation is that yes, there is traffic limitations due to having to compete for bandwidth at peak times, but at non-peak times, I'd rather have a better technical explanation of why this speed can't be achieved.


Have a read of these two articles - they will start giving you the basic idea.

http://en.wikipedia.org/wiki/TCP_tuning
http://cisconet.com/traffic-analysis/throughput/104-tcp-throughput-calculation-formula.html

I'm making some sweeping generalisations when I say this, but in basic terms the longer the path the traffic has to travel (in terms of time - so a higher latency), the lower the theoretical maximum throughput.

So a path to a local server (i.e. at your ISP's core network) is very short, with very few hops or routers to go through to get there - this means the maximum throughput will be far higher than opening a connection to a server on the other side of the world, where traffic has many hops as it jumps from network to network to reach it's destinaction, In this case, the maximum throughput will be far lower.

By multithreading downloads that are coming from a significant distance away, each one can come through at the (low by comparison) maximum attainable speed - the combined speed of them all, however, will be far higher.

This is also a good article which explains in fairly readable language about latency in networks, and would be worth a read.
http://rescomp.stanford.edu/~cheshire/rants/Latency.html

5187 posts

Uber Geek
+1 received by user: 1681


  Reply # 775194 5-Mar-2013 19:27
Send private message

kentnl:
And the point remains; why do I get poorer speeds on a cable connection, with no intrinsic speed limitation, than my ADSL friends, to the same destinations, when ADSL has technical limitations on the medium, and I do not?


In theory, there should be no reason. However, a couple of possibilities are that cable users are in a different bandwidth pool to xDSL (I have absolutely no idea if this is the case for TC); another would be a misconfigured router at your end, perhaps an MTU setting or something similar. Just a couple of possibilities...

36 posts

Geek
+1 received by user: 4


  Reply # 775252 5-Mar-2013 20:39
Send private message

i thought telstraclear say speeds upto 100Mbps? they don't actually say anywhere that you will get 100

23 posts

Geek
+1 received by user: 4


  Reply # 775378 6-Mar-2013 00:52
Send private message

FelixW: i thought telstraclear say speeds upto 100Mbps? they don't actually say anywhere that you will get 100


When you say "Up To", that means implicitly "it will be this some of the time, at least, for some people".

To compare,, when you have a sale and things are "up to 50% off", that means "not everything is 50% off, but some things are".

If you were to have a sale and say "up to 50% off" with nothing more than 10% off, you'd get in trouble with the commerce commission for false advertising.

Maybe TCL can justify this under "Well, some people do get that speed, .... locally".


Just the word "local" is nowhere in their advertising, so its somewhat under the assumption that this means "For websites I actually visit" which largely means "Websites hosted outside New Zealand".

And I have yet to see anything in their TOS/ T&C of this plan that states any such discrepancy is to be expected.

As it is, I have no idea what sort of speeds to expect internationally, and international speed is *all* I care about.


DSL providers are actually somewhat in a safe scenario, because they can just sell their plan as "as fast as it will go", which will limit them to 20M due to the physical medium, so when people see international traffic in relative range of the value they get back from their speed tests, they go "oh right, I'm getting reasonable speeds for my gear, I'm not getting limited by anything other than the wire to my house being slow".


Sure, perhaps this assumption isn't true, but this is the reasoning most people grew up with, their line can only go as fast as it lets them go, and it can't possibly go faster than "X", and any slowness you experience is not your ISPs fault, its just ADSL being fudgy.


But when you're given superior networking hardware and the limitations of ADSL go away, you're then either left concluding that your ISP is screwing you over when you get substantially less than the [ plan + physical design limitation ] speed, well, either that, or you assume your ISP is incompetent and their connection to the rest of the world is likewise limited in bizzare ways.

Sure, maybe there is a technical reason at Telstras End,  but until now, end users have not directly encountered speed problems due to this limitation ( with the exception of "all you can eat" weekends where the whole network runs to a standstill ), its always been "the medium" being the limitation.


Have a read of these two articles - they will start giving you the basic idea.

http://en.wikipedia.org/wiki/TCP_tuning
http://cisconet.com/traffic-analysis/throughput/104-tcp-throughput-calculation-formula.html

I'm making some sweeping generalisations when I say this, but in basic terms the longer the path the traffic has to travel (in terms of time - so a higher latency), the lower the theoretical maximum throughput.

So a path to a local server (i.e. at your ISP's core network) is very short, with very few hops or routers to go through to get there - this means the maximum throughput will be far higher than opening a connection to a server on the other side of the world, where traffic has many hops as it jumps from network to network to reach it's destinaction, In this case, the maximum throughput will be far lower.

By multithreading downloads that are coming from a significant distance away, each one can come through at the (low by comparison) maximum attainable speed - the combined speed of them all, however, will be far higher.

This is also a good article which explains in fairly readable language about latency in networks, and would be worth a read.
http://rescomp.stanford.edu/~cheshire/rants/Latency.html



Thanks RunningMan, this is definitely the sort of information I was trying to find.  I'll have to get my head in that and try understand it a bit better, as it will help me understand the problem at the ISP end a bit better, and hopefully find a logical explanation for why traffic internationally is ridiculously slower than local traffic.

I know for the average person, too much information just leads to confusion, but for me, the more information I can have on the problem, the better I can understand it, and the better I am to make reasonable complaints within reasonable expected limitations of the design. I love having a situation where the only limitation is my own lack of education/intelligence, its much better than a frustrating standoff with a brick wall that just wants to give me the dumbed down version of everything and feeling I'm getting shanked and there's nothing I can do about it =)


3421 posts

Uber Geek
+1 received by user: 410

Trusted

  Reply # 775382 6-Mar-2013 01:17
Send private message

kentnl: Other Flatmate Geek Here.

We've been in contact with customer service with regard to international traffic speeds, and we've been asked to email them trace-routes of specific servers we're having problems with.


This strikes me as odd personally, because the I see these lacks of speed for practically *every* server on the Internet, not just specific ones. Internationally, every single speed test we try has trouble breaking the 10M barrier.

You'll see here, that this server for instance, sees no telstra clear customers even breaking 5M to any of its test servers: http://testmy.net/hoststats/telstraclear_ltd

Historically we got 80M - 100M on a regular basis to the Snap Speed test server, and Telstra Clears Own Speed Test server, but not today.




For TCP traffic it is virtually impossible to get line speed at that rate for anything on the internet (bar maybe Australia) due to the latency from  NZ to practically anywhere. Onbly way you could get there are many open connections and speedtest doesn't emulate that.





23 posts

Geek
+1 received by user: 4


  Reply # 775388 6-Mar-2013 06:41
Send private message

Zeon:

For TCP traffic it is virtually impossible to get line speed at that rate for anything on the internet (bar maybe Australia) due to the latency from  NZ to practically anywhere. Onbly way you could get there are many open connections and speedtest doesn't emulate that.


 

Right, so lets think "future tense", based on what we presently know about internet technology, is there any tech present or future that we know of that could make this better?

 

I mean, you say "its impossible due to distance", however, the aforementioned "testmy.net" service *does* see 100M throughput occasionally from NZ somehow over existing infrastructure.


And testmy.net's tests run on lowly web-browser-only support, not fancy stuff with flash, and they don't as yet have *any* multi-segmented transfer test support.

http://testmy.net/country/nz

So .... what exactly is going on then?

5187 posts

Uber Geek
+1 received by user: 1681


  Reply # 775487 6-Mar-2013 09:39
Send private message

kentnl:
Right, so lets think "future tense", based on what we presently know about internet technology, is there any tech present or future that we know of that could make this better?

No, it's the rules of physics. Have you read the third article I linked to - it probably has the best explanation. Just ignore the fact it was written nearly 20 years ago, the concept still stands.
 
kentnl:
I mean, you say "its impossible due to distance", however, the aforementioned "testmy.net" service *does* see 100M throughput occasionally from NZ somehow over existing infrastructure.

But you don't know who did the test, and where from that got that speed. It may well be a techo at an ISP running it from a server on their core network that is going to (in terms of latency) be far closer to the test server than any real world customer will ever be. You see the odd speed test image jump up with people showing gigabit speeds from a speedtest server - these are generally from someone who works at an ISP that actually hosts the speedtest server on their network. Basically, ignore the very extreme test results that are out there - they are artificial, not real world.

kentnl:
And testmy.net's tests run on lowly web-browser-only support, not fancy stuff with flash, and they don't as yet have *any* multi-segmented transfer test support.

http://testmy.net/country/nz

So .... what exactly is going on then?


Speed test web sites are simply not a reliable and scalable method of measuring network throughput. It's far more complicated than just going to a speed testing web site, and clicking the go button.

When you are getting to high capacity pipes over long distance, a useful analogy may be a motorway (obviously not an NZ motorway, because we'll assume that everyone is travelling at exactly the speed limit!). Build a motorway from Auckland to Wellington. Want to double the throughput? Add twice as many lanes and you will, but the speed limit is still there for each individual lane. The time taken for a vehicle to get from one end to the other stays the same, no matter how wide the motorway gets.

I think the confusion over "I'm not getting the speed my ISP promised" comes from the difference in terminology. A 100Mb/s connection has a throughput of 100 Megabits per second, not a speed of 100Mb/s. You'll get a total throughput of 100Mb/s under good conditions, reasonably often, but the speed of a single TCP connection won't get that high, hence why you need to multithread.

A speed testing site will measure the speed (as it says it will), but may not measure the throughput unless enough threads are open.

But if they used the word throughput, not speed in advertising, most people wouldn't know what it was.


3421 posts

Uber Geek
+1 received by user: 410

Trusted

  Reply # 775490 6-Mar-2013 09:41
Send private message

A good way to test your real speed to destinations is multiple stream Iperf tests. You need an Iperf server to connect to though. For example I was only getting 3mbps from a Stream from Orcon to Los Angeles but using 128 Iperf streams I was getting 270mbps....





1 | 2 | 3
Filter this topic showing only the reply marked as answer View this topic in a long page with up to 500 replies per page Create new topic



Twitter »

Follow us to receive Twitter updates when new discussions are posted in our forums:



Follow us to receive Twitter updates when news items and blogs are posted in our frontpage:



Follow us to receive Twitter updates when tech item prices are listed in our price comparison site:



Geekzone Live »

Try automatic live updates from Geekzone directly in your browser, without refreshing the page, with Geekzone Live now.



Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly to your computer or smartphone by using a feed reader.

Alternatively, you can receive a daily email with Geekzone updates.