Why is Steam uploading whilst i download a game now?

[hex4]

This is a real issue as i have an ADSL2+ connection, 18 down, 797kbps up, i used to get full bandwidth, now i barely get 10 and my ISP does not throttle anything.

 

Uploading kills my connection.

 

Task manager updates the speeds much faster than Steam, also yes i use WiFi but it's literally 3 foot away from the router.

 

 

As you can see i usually throttle it ti 1.5 so that i have some overhead left, even though the upload is not maxed it makes a huge difference on the DL speed.

[Enderman]

248kbps is throttling your download speed??

have you considered maybe its just the server you're downloading from?

[hex4]

1 minute ago, Enderman said:

248kbps is throttling your download speed??

have you considered maybe its just the server you're downloading from?

Both London and Manchester do the same thing, Steam never uploaded before either.

 

Also that spikes above 500 kbps.

[manikyath]

pretty sure thats how file transfer works.

 

in fact, downloading my usual test (latest ubuntu iso) reveals that or the ~20Mbps download speed i get from them, i need around 500kbps upload speed.

this ratio varies on the specific protocols used, but generally is along the lines of the ratio described here.

 

fun fact: those ISPs that do "gigabit download speed" but with something silly like 20Mbps upload, there's no protocol in existance that can make use of that gigabit speed with such a small upload in comparison.

[hex4]

1 minute ago, manikyath said:

pretty sure thats how file transfer works.

 

in fact, downloading my usual test (latest ubuntu iso) reveals that or the ~20Mbps download speed i get from them, i need around 500kbps upload speed.

this ratio varies on the specific protocols used, but generally is along the lines of the ratio described here.

 

fun fact: those ISPs that do "gigabit download speed" but with something silly like 20Mbps upload, there's no protocol in existance that can make use of that gigabit speed with such a small upload in comparison.

Right so around 2 days a go and 6 solid years before hand must have been a dream then, ok.

[manikyath]

1 minute ago, hex4 said:

Right so around 2 days a go and 6 solid years before hand must have been a dream then, ok.

have you considered maybe your upload may be less than usual today?

 

if you dont believe me, this is what a speedtest looks like for me on rainmeter:

[hex4]

1 minute ago, manikyath said:

have you considered maybe your upload may be less than usual today?

 

if you dont believe me, this is what a speedtest looks like for me on rainmeter:

I log right into my router, it's 797 up and 17765 down.

[manikyath]

Just now, hex4 said:

I log right into my router, it's 797 up and 17765 down.

do a speedtest, my router gladly tells me values that arent even sensible

[hex4]

4 minutes ago, manikyath said:

do a speedtest, my router gladly tells me values that arent even sensible

This is from the stickied topic, this is a bad result to a far server.

 

[manikyath]

2 minutes ago, hex4 said:

This is from the stickied topic, this is a bad result to a far server.

 

as in, one done today, since today you're having issues.

[hex4]

1 minute ago, manikyath said:

as in, one done today, since today you're having issues.

This was done a few days a go...

 

[manikyath]

13 minutes ago, hex4 said:

This was done a few days a go...

 

yes, and i ask you for a fresh one because i expect it wont be the same as the one you linked

[hex4]

24 minutes ago, manikyath said:

pretty sure thats how file transfer works.

 

in fact, downloading my usual test (latest ubuntu iso) reveals that or the ~20Mbps download speed i get from them, i need around 500kbps upload speed.

this ratio varies on the specific protocols used, but generally is along the lines of the ratio described here.

 

fun fact: those ISPs that do "gigabit download speed" but with something silly like 20Mbps upload, there's no protocol in existance that can make use of that gigabit speed with such a small upload in comparison.

I feel so fucking retarded, i feel i just wasted every ones time and i should just axe my own head off.

 

Reset the Router channel and now everythinf is swimming fine, checked the upload usage and it's spiking the same as before but DL is maxed out even hitting 18 down.

 

Sorry for my bullshit.

 

 

 

[manikyath]

7 minutes ago, hex4 said:

I feel so fucking retarded, i feel i just wasted every ones time and i should just axe my own head off.

 

Reset the Router channel and now everythinf is swimming fine, checked the upload usage and it's spiking the same as before but DL is maxed out even hitting 18 down.

 

Sorry for my bullshit.

 

 

 

congrats. your modem derped and didnt get its rated upload speed.

 

step 1 when you're having speed issues: speedtest

step 1.5: reset modem

step 2: go back to what you were doing.

[.spider.]

15 hours ago, manikyath said:

 

fun fact: those ISPs that do "gigabit download speed" but with something silly like 20Mbps upload, there's no protocol in existance that can make use of that gigabit speed with such a small upload in comparison.

UDP or proper tuned TCP

[manikyath]

2 minutes ago, .spider. said:

UDP or proper tuned TCP

by default, in a client-server enviroment, where the end user with this type of connection *is* the client, there will always have to be a request coming from the client in order for the server to send data. and when we're talking values where the download speed is 50 times more than the upload speed, theres very little to be done for that upload to not be an absolute bottleneck.

 

look at my earlier posted speedtest result, if there's one place where you can optimize a download, it has to be speedtest right? well, truth of the matter is even speedtest manages a pretty meager 40x download vs upload ratio. the example of the 1000/20 connection only being a 50x ratio, which can result in a very tight scenario.

 

truth is indeed it starts to matter less towards the higher speeds, but that doesnt mean this upload overhead can be forgotten.

[.spider.]

29 minutes ago, manikyath said:

 

truth is indeed it starts to matter less towards the higher speeds, but that doesnt mean this upload overhead can be forgotten.

With UDP it can because there is none

[Yamoto42]

3 hours ago, manikyath said:

by default, in a client-server enviroment, where the end user with this type of connection *is* the client, there will always have to be a request coming from the client in order for the server to send data. and when we're talking values where the download speed is 50 times more than the upload speed, theres very little to be done for that upload to not be an absolute bottleneck.

 

look at my earlier posted speedtest result, if there's one place where you can optimize a download, it has to be speedtest right? well, truth of the matter is even speedtest manages a pretty meager 40x download vs upload ratio. the example of the 1000/20 connection only being a 50x ratio, which can result in a very tight scenario.

 

truth is indeed it starts to matter less towards the higher speeds, but that doesnt mean this upload overhead can be forgotten.

 

...what?  you aren't making a bit of sense...

 

2 hours ago, .spider. said:

With UDP it can because there is none

This on the other hand, perfect sense.

[manikyath]

1 minute ago, Yamoto42 said:

This on the other hand, perfect sense.

if that makes sense to you, please explain me how the server knows it has to send you data?

[Yamoto42]

43 minutes ago, manikyath said:

if that makes sense to you, please explain me how the server knows it has to send you data?

I don't know...maybe use a 4K request packet with a 50MB response?

A 50:1 UL:DL ratio is HARDLY an absolute bottleneck by any stretch of the imagination.

 I'm tempted to not bring up that a TCP ACK can be only 20 bytes, whilst a data payload can reach almost 64KB.  Now realistically your typical MTU will be 1500B...but this is still a 75:1 ratio, and that's not even accounting for the fact that not every TCP packet must be ACKed.  But honestly, after that absolute minimum do we really need to calculate the window?  But as @.spider. mentioned, even that can be avoided by using UDP at layer 4.

But the interesting thing is with that statement you aren't even talking about TCP.  You're talking about an application level protocol that could use a  single TCP packet within that MTU to make a request that gives GIGABYTES of data in response with no further action required by the client.

 

And I'll just add this as a bonus:  Not one bit of evidence has been presented that Steam is the primary source of the outgoing traffic.

The most likely cause of the problem:

 

20 hours ago, Enderman said:

248kbps is throttling your download speed??

have you considered maybe its just the server you're downloading from?

It probably doesn't have anything to do with him, or his computer, or his network.

[manikyath]

1 minute ago, Yamoto42 said:

And I'll just add this as a bonus:  Not one bit of evidence has been presented that Steam is the primary source of the outgoing traffic.

The most likely cause of the problem:

 

It probably doesn't have anything to do with him, or his computer, or his network.

if the guy resetting his modem fixed the issue, how big is the chance that suddenly made the server he's downloading from faster? that'd be one hell of a coincidence would it?

 

also, i provided you with a (somewhat) real world example of what a ratio would be like. if you want to convince me, please come with some decent data on what's possible instead of talking about theory.

 

because lets be honest here, in "theory" a processor with hyperthreading is twice as fast as one without.

[Yamoto42]

If the guy resetting his modem fixed the issue, that does absolutely nothing to show that upload speed is related to the download.  Which is your claim, and thus requires the burden of proof.  I simply showed that it was possible for your claim to be false.