BigPipe's UFB contention ratio.

Forums › Spark New Zealand (including Skinny and BigPipe) › BigPipe's UFB contention ratio.

1 | 2 | 3 | 4

richms

28057 posts

Uber Geek

Trusted

Lifetime subscriber

#1646950 6-Oct-2016 21:45

Imma gonna leave you all here to discuss this stuff.

My answer still stands, bigpipe delivers line speed during peak times with no evidence of a problem with contention.

Richard rich.ms

unsignedint

69 posts

Master Geek

#1646961 6-Oct-2016 21:52

meesham:

Truenet use multiple devices/connections to test with and don't agree with your results... https://www.truenet.nz/websites

Yeah, I'm not even trying to test "website loading performance" but latency. But, if you have a gander at https://www.truenet.nz/international-latency actually you'll find the results do kinda agree.

Sydney: Spark loses to 2degrees, and the worst of all except Vodafone (which looks like something funky is going on with that route)

Melbourne: 2degrees result is a complete outlier; but Spark is the worst of all the others. (A single data point I would guess... *sigh*)

Dallas: Spark are the worst

UK: Spark are the worst

Hmmm... looks kind similar to what I found, but I had 300 test points not 4 :-)

meesham

973 posts

Ultimate Geek

#1646977 6-Oct-2016 22:16

unsignedint:

meesham:

Truenet use multiple devices/connections to test with and don't agree with your results... https://www.truenet.nz/websites

Yeah, I'm not even trying to test "website loading performance" but latency. But, if you have a gander at https://www.truenet.nz/international-latency actually you'll find the results do kinda agree.

Sydney: Spark loses to 2degrees, and the worst of all except Vodafone (which looks like something funky is going on with that route)

Melbourne: 2degrees result is a complete outlier; but Spark is the worst of all the others. (A single data point I would guess... *sigh*)

Dallas: Spark are the worst

UK: Spark are the worst

Hmmm... looks kind similar to what I found, but I had 300 test points not 4 :-)

Unless you're gaming in the US latency tests don't mean much, in my opinion website load speed is more important - and there's not much in it for the latency results to LA anyway.

I'd also argue their testing method for testing latency is superior - they're using multiple connections at different times of the day, they make sure there's no other traffic on the connection while they test and they're averaging over a period of time. If you want to base your decisions on your latency tests that's your business, personally I think you're just trying to spread FUD.

Jase2985

13420 posts

Uber Geek

ID Verified

Lifetime subscriber

#1647016 7-Oct-2016 05:32

unsignedint:

Jase2985:

where are the connections? what devices were used?

im just trying to see if you are comparing apples with oranges or not

You criticised me for making a baseless statement and not providing any data, which was fair enough (though my exact words were "I'm not a fan of.."). So to appease you and not look foolish, I literally:

1. wrote a program to scrape all the Ubuntu CD mirrors (which I think is a fairly good cross-section of the Internet, Ubuntu being the most popular Linux distribution and probably powering half the servers that are online anyway)

2. ran said program to collect data from hosts, connected via Ethernet, that I have access to on those three networks. Vibe and GG were Ubuntu VMs (one at Sitehost, one at my office), and 2degrees was a FreeBSD NAS box residential. I believe the measurements in my results should be what one would expect if they were on a fibre connection to that ISP in Auckland.

3. wrote a program to analyse the results and declare the "worst" if the 10ms difference threshold was exceeded (to once again, account for possible "fuzz" and really if they are all within 10ms I think its fair to say there is no difference).

4. posted results AND the program in public (so anyone can run it and compare with their own results)

Bananas?

Thankfully, the results seems to substantiate my originally baseless claim. I don't have a vested interest in any ISP here. I don't have affiliate links in my signature. So these are the results I found today, it could be completely different tomorrow, but I would suspect not :)

Thank you, all i was trying to assataine was that they were tested via the same type of connection via the same type of device. ie the testing method was the same

for all we know you could have tested ADSL/VDSL VS Fibre

NzBeagle

961 posts

Ultimate Geek

Trusted

#1647021 7-Oct-2016 07:38

unsignedint:

meesham:

Truenet use multiple devices/connections to test with and don't agree with your results... https://www.truenet.nz/websites

Yeah, I'm not even trying to test "website loading performance" but latency. But, if you have a gander at https://www.truenet.nz/international-latency actually you'll find the results do kinda agree.

Sydney: Spark loses to 2degrees, and the worst of all except Vodafone (which looks like something funky is going on with that route)

Melbourne: 2degrees result is a complete outlier; but Spark is the worst of all the others. (A single data point I would guess... *sigh*)

Dallas: Spark are the worst

UK: Spark are the worst

Hmmm... looks kind similar to what I found, but I had 300 test points not 4 :-)

Need more BigPipe volunteers to sign up, but they're not even looking for any according to the website...

Yabanize

2350 posts

Uber Geek

#1647970 9-Oct-2016 13:27

NzBeagle:

unsignedint:

meesham:

Truenet use multiple devices/connections to test with and don't agree with your results... https://www.truenet.nz/websites

Yeah, I'm not even trying to test "website loading performance" but latency. But, if you have a gander at https://www.truenet.nz/international-latency actually you'll find the results do kinda agree.

Sydney: Spark loses to 2degrees, and the worst of all except Vodafone (which looks like something funky is going on with that route)

Melbourne: 2degrees result is a complete outlier; but Spark is the worst of all the others. (A single data point I would guess... *sigh*)

Dallas: Spark are the worst

UK: Spark are the worst

Hmmm... looks kind similar to what I found, but I had 300 test points not 4 :-)

Need more BigPipe volunteers to sign up, but they're not even looking for any according to the website...

Yeah, I tried to sign up to Truenet as a Bigpipe Christchurch UFB tester but they weren't interested.

Jase2985

13420 posts

Uber Geek

ID Verified

Lifetime subscriber

#1647972 9-Oct-2016 13:29

they probably have enough testers at the moment, they are only funded for so many per ISP

Quic Broadband

Move to New Zealand's best fibre broadband service (affiliate link). Note that to use Quic Broadband you must be comfortable with configuring your own router.

Yabanize

2350 posts

Uber Geek

#1647974 9-Oct-2016 13:35

Jase2985:

they probably have enough testers at the moment, they are only funded for so many per ISP

Possibly, but I signed up very shortly after Bigpipe made UFB available in Christchurch

Jase2985

13420 posts

Uber Geek

ID Verified

Lifetime subscriber

#1647980 9-Oct-2016 13:47

but again they probably have enough testest for UFB at present. they only get funding for x number per ISP. Bigger ISP's gets more testers.

dont forget those who may have been testers for other ISP's that have switched to bigpipe are probably still testers for truenet

happyfunball

287 posts

Ultimate Geek

#1655654 21-Oct-2016 12:45

meesham:

Unless you're gaming in the US latency tests don't mean much, in my opinion website load speed is more important - and there's not much in it for the latency results to LA anyway.

I'd also argue their testing method for testing latency is superior - they're using multiple connections at different times of the day, they make sure there's no other traffic on the connection while they test and they're averaging over a period of time. If you want to base your decisions on your latency tests that's your business, personally I think you're just trying to spread FUD.

You have that backwards. website load speeds don't mean much, because web pages can be cached on multiple levels, including by the ISP and CDN's to cut down on international bandwidth usage.

IMO in this age of unlimited data, latency is the best measure of network connection quality.

1 | 2 | 3 | 4