Why is my ping and upload speed so bad despite the fact I have great download speed

[PutinCare]

So I recently built a new computer for Christmas and It's downloads speeds are great and I normally get between 5Mb/s - 9Mb/s (on steam). but when I went into TF2 I was getting ping of 300 on a Sydney server (I live in Melbourne BTW). also I have a terrible upload speed. At the moment I'm using a pretty high end Network card (http://www.msy.com.au/wireless/15399-tp-link-archer-t9e-ac1900-wireless-dual-brand-pci-e-adapter.html) as using an Ethernet cable is not possible for me at the moment. anyone know how to fix this. (also yes I am close to the router)  

[Guest]

Only Sydney server? Any high ping anywhere else? OR just that?

[PutinCare]

2 minutes ago, deXxterlab97 said:

Only Sydney server? Any high ping anywhere else? OR just that?

I said Sydney because that's the only server in Australia, everywhere else is worse 

[Guest]

1 minute ago, PutinCare said:

I said Sydney because that's the only server in Australia, everywhere else is worse 

only tf2? or any other game? have you contacted ISP for help?

[iRileyx]

does steam show MBps or Mbps when downloading.

 

also try running speed test and post the result. But its quite normal for upload to be alot less than download, I can get 150 down and only 10 up at times

[PCGeek]

Try to run a tracert to see where the bottleneck is.  Also check if there is any funky MTU sizing going on.

[PCGeek]

3 minutes ago, PutinCare said:

I said Sydney because that's the only server in Australia, everywhere else is worse 

Melbourne to Sydney should only be around 20ms.

[iRileyx]

2 minutes ago, PCGeek said:

Melbourne to Sydney should only be around 20ms.

Did google tell you that?

[PCGeek]

1 minute ago, RKRiley said:

Did google tell you that?

No just an estimate since it's about 500 miles.  Actually should be even less.

[iRileyx]

3 minutes ago, PCGeek said:

No just an estimate since it's about 500 miles.  Actually should be even less.

surely it also depends on the isp, the load on the isp servers, and (if the ping was read in tf2) the load on the tf2 servers?

 

seems quite hard to guess with all the factors.

[Ryoku]

3 minutes ago, PCGeek said:

No just an estimate since it's about 500 miles.  Actually should be even less.

The internet doesn't necessarily go in a straight line, it can go like 500 mile circle around before getting to its destination making it 1000 mines instead (just an example).  Routing can just plain be bad.

[Lyrin]

DEPENDING on how much traffic there is, you're looking somewhere between 10ms > 50ms average with the distance, but I don't think that is the case here.

 

Could be bad port forwarding, could be your wireless is conflicting with other local Wireless networks (use a tool to see how many local networks are using your channel. Example for Android phones: Wifi Analyser.) Could be peak hour and everyone is using the net.

 

Or it could be issues with your ISP and how they have set things up.

[PutinCare]

Ok so I did a speed test from the closest server In Melbourne to Sydney and this is what I got

[PCGeek]

4 minutes ago, Ryoku said:

The internet doesn't necessarily go in a straight line, it can go like 500 mile circle around before getting to its destination making it 1000 mines instead (just an example).  Routing can just plain be bad.

Absolutely correct but generally you follow submarine cable maps and such.  Melbourne to Sydney is generally 800 miles based on the M31 highway where the cables are laid.

[PCGeek]

8 minutes ago, RKRiley said:

surely it also depends on the isp, the load on the isp servers, and (if the ping was read in tf2) the load on the tf2 servers?

 

seems quite hard to guess with all the factors.

Well if it is then the I would say get a different ISP.  Generally in this day and age there shouldn't be a huge variance.  If you follow the lines that cables are laid be it over land or over submarine cables there shouldn't be much more extra than that.  If it is then the ISPs are using poor peering methods.

[PCGeek]

12 minutes ago, RKRiley said:

surely it also depends on the isp, the load on the isp servers, and (if the ping was read in tf2) the load on the tf2 servers?

 

seems quite hard to guess with all the factors.

I once had a customer in Philippines who were using a peering arrangement where they were routing traffic through NTT Data in Japan before getting the traffic to Hong Kong where they could have routed it directly from Philippines in Batangas to Hong Kong.  I requested they make this change and the latency went from >90ms to 24ms.

[LAwLz]

7 hours ago, PCGeek said:

Melbourne to Sydney should only be around 20ms.

What makes you think that?

You can't just look at the distance and go "oh it should be this latency". There are far more variables at play.

[PCGeek]

9 minutes ago, LAwLz said:

What makes you think that?

You can't just look at the distance and go "oh it should be this latency". There are far more variables at play.

Not quite, you look at the way the submarine cables or over land cables are laid. In the early days of the Internet there were fewer POPs (Points of Presence) so it was harder to determine but nowadays you should be able to determine by distance in first world countries since there are more POPs.  If the latency is higher than that it means the ISP has crappy peering or inefficient routing and in this day and age there's no excuse for that.  I would change ISPs if that was the case.  Attached are my ping times from Sydney to Melbourne. It's actually better than I estimated by 5ms.

[LAwLz]

4 minutes ago, PCGeek said:

Not quite, you look at the way the submarine cables or over land cables are laid. In the early days of the Internet there were fewer POPs (Points of Presence) so it was harder to determine but nowadays you should be able to determine by distance in first world countries since there are more POPs.  If the latency is higher than that it means the ISP has crappy peering or inefficient routing and in this day and age there's no excuse for that.  I would change ISPs if that was the case.  Attached are my ping times from Sydney to Melbourne. It's actually better than I estimated.

Again, there are far more variables than that. You're dramatically oversimplifying things and while you might be right sometimes, you might also be very off. Like your own numbers show, you were over 40% off.

 

How did you estimate the latency? Because you can't just look at the cables and then use some static speed like "speed of light in a vacuum", because that does not take into consideration any encoding time, processing time, signal amplification, potential detours or delays, or any of the other long list of things which all adds up to the final latency.

 

Hell, we don't even know if OP has fiber, ADSL, DOCSIS or some other type of internet connection, and all of those has very different processing times.

Not to mention that OP is on wireless, so that might play a major role in determining the latency too, depending on how clean and uninterrupted that signal is.

[PCGeek]

5 minutes ago, LAwLz said:

Again, there are far more variables than that. You're dramatically oversimplifying things and while you might be right sometimes, you might also be very off. Like your own numbers show, you were over 40% off.

 

How did you estimate the latency? Because you can't just look at the cables and then use some static speed like "speed of light in a vacuum", because that does not take into consideration any encoding time, processing time, signal amplification, potential detours or delays, or any of the other long list of things which all adds up to the final latency.

 

Hell, we don't even know if OP has fiber, ADSL, DOCSIS or some other type of internet connection, and all of those has very different processing times.

Not to mention that OP is on wireless, so that might play a major role in determining the latency too, depending on how clean and uninterrupted that signal is.

I was off by 25%.  I estimated 20ms and it ended up being 15ms.  It's only a baseline.  Yes WiFi can add some additional latency but you need to have a baseline to tune towards.  Also fiber, ADSL, DOCSIS will generally affect bandwidth numbers but latency shouldn't be too far off.  When we look at variances we don't want to be off by much more than 10ms as a general practice.  The OP was hitting over 300ms so we know there is definitely an issue.  As for the estimates it considers more than just speed of light.  It's a practical estimate from experience.

[LAwLz]

1 minute ago, PCGeek said:

Also fiber, ADSL, DOCSIS will generally affect bandwidth numbers but latency shouldn't be too far off.

But it is... Again, the processing required to actually encode and decode the signal is widely different between the two, and thus they do have a major impact on the latency.

 

 

8 minutes ago, PCGeek said:

The OP was hitting over 300ms so we know there is definitely an issue.

I agree with that.

 

8 minutes ago, PCGeek said:

As for the estimates it considers more than just speed of light.  It's a practical estimate from experience.

For example?

[PCGeek]

13 minutes ago, LAwLz said:

But it is... Again, the processing required to actually encode and decode the signal is widely different between the two, and thus they do have a major impact on the latency.

 

 

I agree with that.

 

For example?

encoding and decoding the signal should not have a huge impact as most of the resequencing issues were fixed in DOCSIS 3.0.  Again, I mentioned that if there is issues then the ISP needs to fix their crap because there is no excuse for that today.

 

My experience generally takes all the other factors you mentioned into consideration.  I'm a Cloud Architect for Microsoft so on our Edge Network, and other Content Delivery networks, we generally have to take a look at expected latency and we have numbers for certain distances. For example from Vancouver to Toronto it's about 50ms of latency at a distance of 2200 miles.  I generally use that as a baseline when estimating latency in North America.  In APAC I measure it by looking at submarine cable maps.  The goal is to use these experiences to generate a baseline.  If there are other things at play that adversely impact latency we try to remediate or at least record it.  Also latency is limited by distance since you can't travel faster than the speed of light.  So much emphasis is really on latency since this is one of the most user experience impacting metrics.  Even more so than bandwidth a lot of the time.

[LAwLz]

2 hours ago, PCGeek said:

encoding and decoding the signal should not have a huge impact as most of the resequencing issues were fixed in DOCSIS 3.0.  Again, I mentioned that if there is issues then the ISP needs to fix their crap because there is no excuse for that today.

Ehm... First of all, there is way more to encoding and decoding the signal than just re-sequencing issues. Secondly, a change to the DOCSIS spec won't help on an xDSL connection.

 

2 hours ago, PCGeek said:

My experience generally takes all the other factors you mentioned into consideration.

Can you please give me a formula for how you calculated what latency OP should have? If you're claiming that you're taking all those other factors into consideration then you must have some pretty complicated formula.

 

2 hours ago, PCGeek said:

Also latency is limited by distance since you can't travel faster than the speed of light.

Well of course. It's limited how long the latency can be, but not how high.

 

The problem is that by saying something like "you should only have 20ms" when you know next to nothing about OP can be extremely misleading. Not only will he think something is majority wrong even if it turned out his latency was just fine, but other people seeing that post might also get the wrong impression.

We have no idea what latency OP should be expecting. All we can say is that 300ms seems very high and he should try and lower it, but saying that he should be getting 20ms can be very misleading.

 

For example if OP is sitting on a xDSL line which has problem with noise, then his ISP might be using a high interleave setting to counter that, as without it even a small amount of noise burst could cause the OP to completely lose his connection. You can't say the ISP is in the wrong for caring more about having a stable connection, rather than one with low latency.

That's just one of the many examples of why different connection types can have different latency. There are many, many, many more reasons why your latency might be different, that isn't "because you're X miles away".