Little difference between providers

Forums › New Zealand Broadband › Little difference between providers

1 | 2 | 3 | 4

Wombat1

586 posts

Ultimate Geek
Inactive user

#3103226 12-Jul-2023 14:30

nitro:

as for the plane trip analogy, a better question to ask would be whether an Airbus A380 or an SR71 would be faster at getting 500 guests from Auckland to attend a wedding in Queenstown.

Valid question, but we not talking about the max amount of data possible, ie 500 guests. Lets say we just want 2 guests. Hence the reason this op is saying there is little difference between providers. All valid points none the less.

Which ISP would download a 25 mb file faster?

ISP 1 (1000/200) on cable or sattelite with around 60ms of latency from the source of data (VodafonesFibreX plan comes to mind here)

ISP 2 (50/20) on fibre with 2ms of latency to the source of data .

I could be wrong, but I would not put my bets on ISP1

Wombat1

586 posts

Ultimate Geek
Inactive user

#3103229 12-Jul-2023 14:46

I have also seen some ISP's throwing around the analogy of a hosepipe, bigger hosepipes can get more water through it and faster blah blah blah... while this is true for a hosepipe, it does not stand true for broadband. Light is already traveling at max speed through the fibre no matter how much wider you make the pipe.

Wheelbarrow01

1714 posts

Uber Geek

Trusted

Chorus

#3103482 12-Jul-2023 22:18

Wombat1:

I have also seen some ISP's throwing around the analogy of a hosepipe, bigger hosepipes can get more water through it and faster blah blah blah... while this is true for a hosepipe, it does not stand true for broadband. Light is already traveling at max speed through the fibre no matter how much wider you make the pipe.

Hmm I'd argue the speed of the water travelling through each hose is about the same, but you end up with more water out of the bigger hose over any given time period. It might seem like it's traveled faster out of the bigger hose but it hasn't - there's just more of it.

The same for data transfer I would have thought. Mbps isn't measuring how fast the data is travelling per second, it's measuring how much data can travel in that second.

If we measured highway limits the same way we measure internet 'speed', the limit over the harbour bridge could be 10000 cars per hour peak time. If you add two more lanes (bigger bandwidth) the cars per hour limit might increase to 12500, but does the speed of each car actually increase? Perhaps not.

The views expressed by me are not necessarily those of my employer Chorus NZ Ltd

nitro

648 posts

Ultimate Geek

#3103513 13-Jul-2023 08:17

Wombat1:

Valid question, but we not talking about the max amount of data possible, ie 500 guests. Lets say we just want 2 guests. Hence the reason this op is saying there is little difference between providers. All valid points none the less.

Which ISP would download a 25 mb file faster?

ISP 1 (1000/200) on cable or sattelite with around 60ms of latency from the source of data (VodafonesFibreX plan comes to mind here)

ISP 2 (50/20) on fibre with 2ms of latency to the source of data .

I could be wrong, but I would not put my bets on ISP1

in this example (perhaps, you mean 25 MB file?), there really would be 'little difference'. you said so yourself with your own experience having used 50/10 and 1000/200 services.

many factors affect latency, e.g. how it's routed from one point to the other, the devices it has to pass through, congestion at any point/s along the path, a bunch of other stuff here.

but as Talkiet and Wheelbarrow01 have said, internet services (particularly residential ones) are generally about the amount of data (megabits) the connection can move per unit of time (seconds). so internet speed is sold per the rate of data transfer.

if you're not hitting the limit, there's going to be 'little difference'.

but if you do... a 50/10 Mbps, 0 ms latency connection is not going to be better than a 1000/200 Mbps 150 ms connection for downloading a 100 MB file. this on a like-for-like comparison of fibre vs fibre, as other methods are more likely to introduce artefacts like dropped/errored packets.

seriously, if you're sitting in front of your pc and noticing the difference in download times of a 25 MB file between a 1000 Mbps and a 50 Mbps internet connections... i suggest we go out and i'll buy you a beer/whisky/coffee or whatever your beverage of choice is. 🥃😊

cddt

1513 posts

Uber Geek

#3103514 13-Jul-2023 08:19

Wombat1:

Most Mbps you will ever get out of any ADSL connection is 24/2.5. And latency is very very bad

Remember when ADSL latency was considered fast... I finally convinced my parents to upgrade to ADSL (JetStart = 128 kbps symmetrical) in 2002 and the latency to local servers was less than half that of dialup.

Wombat1

586 posts

Ultimate Geek
Inactive user

#3103538 13-Jul-2023 09:27

nitro:

but if you do... a 50/10 Mbps, 0 ms latency connection is not going to be better than a 1000/200 Mbps 150 ms connection for downloading a 100 MB file. this on a like-for-like comparison of fibre vs fibre, as other methods are more likely to introduce artefacts like dropped/errored packets.

seriously, if you're sitting in front of your pc and noticing the difference in download times of a 25 MB file between a 1000 Mbps and a 50 Mbps internet connections... i suggest we go out and i'll buy you a beer/whisky/coffee or whatever your beverage of choice is. 🥃😊

Which connection would you prefer for gaming? Also if you using a VOIP phone you are far more likely to notice the lag on that 150ms connection when talking to people further away.

Latency has a major impact on download speed because of the way that the TCP protocol works. TCPIP "send window" is set differently between higher and lower latency connections. It will be interesting to do some real world tests. There are some formuals available to use to calculate the download "speed", but none of them that I have seen take into account the latency of the network. If you cut the latency in half, you can probably double your transfer rate (provided that the connection has not reached it max mbps).

Talkiet

4787 posts

Uber Geek

Trusted

#3103546 13-Jul-2023 09:40

Wombat1:

[snip]There are some formuals available to use to calculate the download "speed", but none of them that I have seen take into account the latency of the network. If you cut the latency in half, you can probably double your transfer rate (provided that the connection has not reached it max mbps).

You've not looked very hard at all then... There's a very well known formula for this.

https://lmgtfy.app/?q=bandwidth+delay+product

Cheers - N

Please note all comments are from my own brain and don't necessarily represent the position or opinions of my employer, previous employers, colleagues, friends or pets.

GoodSync

GoodSync. Easily back up and sync your files with GoodSync. Simple and secure file backup and synchronisation software will ensure that your files are never lost (affiliate link).

deadlyllama

1259 posts

Uber Geek

Trusted

#3103573 13-Jul-2023 10:20

Talkiet:

Wombat1:

[snip]There are some formuals available to use to calculate the download "speed", but none of them that I have seen take into account the latency of the network. If you cut the latency in half, you can probably double your transfer rate (provided that the connection has not reached it max mbps).

You've not looked very hard at all then... There's a very well known formula for this.

https://lmgtfy.app/?q=bandwidth+delay+product

Cheers - N

That's a metric, not a speed. High bandwidth delay produce networks have poor TCP throughput with traditional (cubic, reno) TCP congestion control algorithms. If the sender is using a better congestion control algorithm, like BBR, you can get better speeds on high BDP links.

Edit - I should have mentioned that of course usually the sender is someone else, and you have no control over their TCP stack. But if the sender is e.g. a virtual server of yours, far away in Europe, you can play with congestion control, and it really does help, and I've done it.

nitro

648 posts

Ultimate Geek

#3103619 13-Jul-2023 11:34

Wombat1:

Which connection would you prefer for gaming? Also if you using a VOIP phone you are far more likely to notice the lag on that 150ms connection when talking to people further away.

the discussion was about downloading a file, was it not? also, i'd wager most people, including you, won't mind a 150ms lag on VOIP calls. again, there are other things to consider about provisioning VOIP services than just latency. people working in telcos here might want to elaborate.

Wombat1:

Latency has a major impact on download speed because of the way that the TCP protocol works. TCPIP "send window" is set differently between higher and lower latency connections. It will be interesting to do some real world tests. There are some formuals available to use to calculate the download "speed", but none of them that I have seen take into account the latency of the network. If you cut the latency in half, you can probably double your transfer rate (provided that the connection has not reached it max mbps).

it's really not as simple as basic TCP/IP. to get to the other end of the connection the packets will traverse using different methods, e.g. MPLS.

and your proposition that cutting the latency in half will double transfer rate goes out the window when the connection is rate limited.

you might have the idea that i don't understand or care about latency, but i do. as hinted previously i play with these things on a regular basis. being on the vendor side, we actually developed and successfully marketed low latency products (wireless) for specific applications where we measured (not calculated) site latency in terms of nanoseconds (end-to-end in microseconds).

Wombat1

586 posts

Ultimate Geek
Inactive user

#3103636 13-Jul-2023 12:35

nitro:

the discussion was about downloading a file, was it not?

No, its acually about their being little differences between providers which is very true for the average user if all of them are on fbre. Very little difference too for an average user like the op if he was on 50/20 or 1000/200.

nitro:

and your proposition that cutting the latency in half will double transfer rate goes out the window when the connection is rate limited.

Read what I wrote again. You obviously misread it.

1 | 2 | 3 | 4