1GB Internet Very Inconsistent

[Makaveli]

So my ASUS router would not give the 1Gbps throughput with 2x1GHz CPU. However, I was able to get close to the 1Gbps by turing on CTF(Cut-Through-Forwarding) on the ASUS router. I'm not sure what TP-Link calls it, or if they even have an option to enable CTF. I guarantee the Comcast modem has CTF enabled by default, which is why I can do the 1Gbps.

But I ended up building a pfSense router that can easily handle 1Gbps, because I didn't want to leave my security in Comcast's hands.



It's more reliable that speedtest.net for fast connection. And by fast connections I mean near 1Gbps. The reason being is that some of the speedtest.net servers are only 1Gbps, so the results you get can vary greatly if you are on a 1Gbps server and it has some load from other users. The fast.com servers are all 10Gbps AFAIK. So you are less likely to hit an overloaded server that can't provide 1Gbps when testing a 1Gbps connection.

However, after playing around with speedtests over the past week or so, I find the DSLreports speed test to be more accurate than fast.com for 1Gbps connections, and even slower ones.




I've got a symtrical 1Gbps cable connection at my work office. There's really no legitimate reason a home user needs more than 40Mbps upload right now. Hell, most would even notice the difference between 10Mbps up and 100Mbps up.

Here is my fast.com result.



I'm on a Fiber connection 1Gbps down and 750Mbps up.

And I happen to be on a R7000 router running Asus Merlin firmware cpu is 2x1Ghz and CTF is on.

Speedtest.net app results

 

[newtekie1]

Here is my fast.com result.



I'm on a Fiber connection 1Gbps down and 750Mbps up.

And I happen to be on a R7000 router running Asus Merlin firmware cpu is 2x1Ghz and CTF is on.

Speedtest.net app results

Yeah, I never said it was perfect, just more consistent in my experience.

 

[Makaveli]

Yeah, I never said it was perfect, just more consistent in my experience.

From all the post i've seen on the internet fast.com isn't the most reliable however I cannot speak for your experience.

DSL Reports for me is only slightly better and the speedtest app gives me the most consistent results.

 

[TheLostSwede]

I've got a symtrical 1Gbps cable connection at my work office. There's really no legitimate reason a home user needs more than 40Mbps upload right now. Hell, most would even notice the difference between 10Mbps up and 100Mbps up.

Right, because no-one works from home and needs to upload large files... Anything below 100Mbps upload imho is not good enough. If you want to keep using crappy internet providers, please go ahead, but I for one, would prefer to get something better. Sadly I'm stuck with a cable company, so I can't get good upload, regardless if I want to or not.

Also, this discussion was kind of closed and done with, as the OP solved his issue.

And no, TP-Link doesn't appear to have CTF. It seems to be a Broadcom feature though, hence why it might not be on the C7 that the OP used.

Opening...

emulator.tp-link.com

 

[Bill_Bright]

From all the post i've seen on the internet fast.com isn't the most reliable

And what have you seen is? Got a link to the review/study/report?

Not sure what you mean by reliable either. Frankly, I don't see how any could be wrong - assuming you can connect to the testing site. It does not take any special hardware, coding, protocol, or knowledge to count time between ends and then calculate that time into a speed. I mean a bit is a bit and a second is a second.

I like Speedtest.net because I can select the server I want to test against. So I always pick the Speedtest.net server in Washington DC because it just happens to be 1000 miles away from here. That's arbitrary, but since I always select that one, its consistent for me.

But what I cannot control, nor can Speedtest.net or Fast.com or any other speed test site, is the path each test will take. The number of hops and which servers are hit are determined by the Internet at that particular point in time. Nor can we control the amount of congestion or other factors that affect latencies at that specific point in time. Even within your own home network, there are variables that can affect over all speeds. The only factor that is truly consistent (but not really) is between your home and your PoP (point of presence) - the physical location where your ISP connects you to the Internet backbone.

I just tested with Speedtest.net, Fast.com, ATT, Cox (my ISP) and Speakeasy. The results for downloads were 179.5, 180, 176.38, 177.3, 173.9, and for uploads they were 9.8, 10.2, 10.5, 10.3 (Fast does not show upload speeds ).

One to Washington DC, two went to Dallas, Tx but I don't know if the same place. Another went to KCMO, my ISP went Omaha, Nebraska, about 10 miles north of here. IMO, those are pretty consistent.

Sadly I'm stuck with a cable company, so I can't get good upload, regardless if I want to or not.

Why not? Cable is not the restriction. Your ISP and how they decide the upload speeds you get may be. But that's a limit due to someone making that decision. As a technology, cable can easily deliver upload speeds equal to download speeds - if the ISP wanted to.

 

[Makaveli]

And what have you seen is? Got a link to the review/study/report?

Not sure what you mean by reliable either. Frankly, I don't see how any could be wrong - assuming you can connect to the testing site. It does not take any special hardware, coding, protocol, or knowledge to count time between ends and then calculate that time into a speed. I mean a bit is a bit and a second is a second.

I like Speedtest.net because I can select the server I want to test against. So I always pick the Speedtest.net server in Washington DC because it just happens to be 1000 miles away from here. That's arbitrary, but since I always select that one, its consistent for me.

But what I cannot control, nor can Speedtest.net or Fast.com or any other speed test site, is the path each test will take. The number of hops and which servers are hit are determined by the Internet at that particular point in time. Nor can we control the amount of congestion or other factors that affect latencies at that specific point in time. Even within your own home network, there are variables that can affect over all speeds. The only factor that is truly consistent (but not really) is between your home and your PoP (point of presence) - the physical location where your ISP connects you to the Internet backbone.

I just tested with Speedtest.net, Fast.com, ATT, Cox (my ISP) and Speakeasy. The results for downloads were 179.5, 180, 176.38, 177.3, 173.9, and for uploads they were 9.8, 10.2, 10.5, 10.3 (Fast does not show upload speeds ).

One to Washington DC, two went to Dallas, Tx but I don't know if the same place. Another went to KCMO, my ISP went Omaha, Nebraska, about 10 miles north of here. IMO, those are pretty consistent.
Why not? Cable is not the restriction. Your ISP and how they decide the upload speeds you get may be. But that's a limit due to someone making that decision. As a technology, cable can easily deliver upload speeds equal to download speeds - if the ISP wanted to.

What I've seen on multiple forums from multiple different users on different ISP posting results. I'm not going to go and collect them all and provide links to prove my point.

And I agree with your point users don't have control over which peering their ISP's use or the other variables you have posted so no disagreement there.

However you are wrong about Cable and its uploads.

"As a technology, cable can easily deliver upload speeds equal to download speeds - if the ISP wanted to."

Umm no if they could they would this is not possible on docsis 3.1 I suggest reading more about Docsis networks before making statements like this.

Cable uploads have been an issue for along time, the someone making the decision on the ratio of the splits is force to do so based on the network. The current version of Docsis 3.1 just doesn't provide enough upload.

I was a cable customer for 15 years so i've been through Docsis 1.0, 2.0, 3.0, 3.1 and now i'm on fiber, and you could not pay me to go back to cable its just inferior technology.

There is hope for Cable but it looks to be years from now.

CableLabs Kicks Off Pursuit of DOCSIS 4.0 | Light Reading

New specs will repackage recent annexes and enhancements to DOCSIS 3.1, such as Full Duplex DOCSIS and Low Latency DOCSIS, and add in new Extended Spectrum DOCSIS capabilities.

www.lightreading.com

 

[silentbogo]

Not sure what you mean by reliable either. Frankly, I don't see how any could be wrong - assuming you can connect to the testing site. It does not take any special hardware, coding, protocol, or knowledge to count time between ends and then calculate that time into a speed. I mean a bit is a bit and a second is a second.

Just like with speedtest, there are many factors that can skew the result. Mainly: server location(s), routing, relative load etc.
We've had similar discussion about Speedtest.net somewhere here on the forum. I think the most reliable way is either manual speedtest to the nearest hand-picked well-known server with huge bandwidth or do the same for iperf.
I used to test our work servers for reporting purposes with speedtest CLI, but now I do both iperf3 and speedtest. Iperf is more consistent, cause it's harder for mortals to use, which means public servers are always empty.

fast.com shows pretty accurate result for me, but it's mostly due to the fact that Netflix barely started operating in Ukraine this year and fast.com testing service is not very popular. Pings are still awful, cause it's friggin' AWS (and routed through Warsaw and Budapest, which for some weird reason always have worse pings than Frankfurt or even Amsterdam).

 

[newtekie1]

Right, because no-one works from home and needs to upload large files... Anything below 100Mbps upload imho is not good enough. If you want to keep using crappy internet providers, please go ahead, but I for one, would prefer to get something better. Sadly I'm stuck with a cable company, so I can't get good upload, regardless if I want to or not.

At that point you aren't really a home user, you're a business user, get a business connection if you want. But for home use, 40Mbps is more than enough.

Umm no if they could they would this is not possible on docsis 3.1 I suggest reading more about Docsis networks before making statements like this.

DOCSIS 3.1 can do 1000/1000 symmetric. It's just not something most home ISPs will give because it isn't necessary for the large majority of home users.

 

[advanced3]

Just hope you live in area that doesn't let Comcast enforce their 1Tb a month usage limit.

 

[Makaveli]

At that point you aren't really a home user, you're a business user, get a business connection if you want. But for home use, 40Mbps is more than enough.




DOCSIS 3.1 can do 1000/1000 symmetric. It's just not something most home ISPs will give because it isn't necessary for the large majority of home users.

Yes it can do 1000/1000 symmetric.

However currently what is offered is 3.1 on the downstream and 3.0 on the upstream.

No one has released 3.1 on upstream.

When that is released I think realistically you will be looking at 100Mbps uploads on current Gbps plans.

You will need Full duplex cable which is I think a few years out and possibly Docsis 4.0.

 

[Bill_Bright]

What I've seen on multiple forums from multiple different users on different ISP posting results.

I would not put much faith in any of those either. Forum posters typically don't have the necessary skills, test equipment, or evaluation experience to make valid tests and comparisons. Their results are anecdotal and little more. That's why I asked for a link to a "review/study/report" - suggesting the comparison was done by someone with some expertise and the necessary tools.

I suggest reading more about Docsis networks before making statements like this.

Gee whiz. I sure don't understand why so many on this site are in such a hurry to launch personal criticisms instead of just debating the technical facts.

Of course it is possible with 3.1 - as you yourself indicated right after newtekie pointed it out! A quick Google search and "reading more about DOCSIS networks before making statements" would easily show that just because the earlier DOCSIS protocols limit upstream bandwidth, that does NOT indicate a limitation in "cable" - which is what my point was in the first place.

DOCSIS 3.1 already supports upsteam rates of "2"Gbps! And as you later noted too, DOCSIS 4.0 supports "full duplex" up to 10Gbps both ways! And nobody is going to have to have their current cable drop to their house replaced just to support DOCSIS 4.0. They will just need an upgraded modem - and a faster home network too - but that's a different discussion.

The fact most ISP don't currently provide it is NOT because "cable" does not support it. It is because the demand for it is not big enough - yet. And it may never happen since the vast majority of home, non-commercial users just don't need that much upstream bandwidth.

 

[newtekie1]

Just hope you live in area that doesn't let Comcast enforce their 1Tb a month usage limit.

It's only $15 a month to get unlimited with Comcast if you are in one of the areas with the 1TB data limit.

 

[Athlonite]

Just hope you live in area that doesn't let Comcast enforce their 1Tb a month usage limit.

most avg users wouldn't go over that in a month anyways and if you do on a regular basis then it's not the plan for you you need to go for an unlimited data plan instead

 

[Aquinus]

It's only $15 a month to get unlimited with Comcast if you are in one of the areas with the 1TB data limit.

You could always move to the glorious state of New Hampshire. Live free or die.

Also, I think Comcast's gigabit pro tier isn't included on the 1TB data usage cap.

 

[advanced3]

most avg users wouldn't go over that in a month anyways and if you do on a regular basis then it's not the plan for you you need to go for an unlimited data plan instead

I regularly pass the 1Tb cap per month (1.5-2Tb) , but I live in Massachusetts so there isn't one in place here. The service worked great for the first month or so, Now for the past 3 months starting around 9pm I get severe packet loss in any game I play, to the point that its unplayable.

Here's a video I uploaded pinging youtube when it starts happening.

 

[Durvelle27]

So my ASUS router would not give the 1Gbps throughput with 2x1GHz CPU. However, I was able to get close to the 1Gbps by turing on CTF(Cut-Through-Forwarding) on the ASUS router. I'm not sure what TP-Link calls it, or if they even have an option to enable CTF. I guarantee the Comcast modem has CTF enabled by default, which is why I can do the 1Gbps.

But I ended up building a pfSense router that can easily handle 1Gbps, because I didn't want to leave my security in Comcast's hands.



It's more reliable that speedtest.net for fast connection. And by fast connections I mean near 1Gbps. The reason being is that some of the speedtest.net servers are only 1Gbps, so the results you get can vary greatly if you are on a 1Gbps server and it has some load from other users. The fast.com servers are all 10Gbps AFAIK. So you are less likely to hit an overloaded server that can't provide 1Gbps when testing a 1Gbps connection.

However, after playing around with speedtests over the past week or so, I find the DSLreports speed test to be more accurate than fast.com for 1Gbps connections, and even slower ones.




I've got a symtrical 1Gbps cable connection at my work office. There's really no legitimate reason a home user needs more than 40Mbps upload right now. Hell, most would even notice the difference between 10Mbps up and 100Mbps up.

Nope my router doesn’t have that setting

At that point you aren't really a home user, you're a business user, get a business connection if you want. But for home use, 40Mbps is more than enough.




DOCSIS 3.1 can do 1000/1000 symmetric. It's just not something most home ISPs will give because it isn't necessary for the large majority of home users.

40Mbps sucks honestly. Really wish I had more for steaming

It's only $15 a month to get unlimited with Comcast if you are in one of the areas with the 1TB data limit.

Where I live I was charged $50 extra for unlimited data when using my own modem.



But to everyone, I plan to switch services as every since I upgraded my speeds to 1000Mbps my internet has sucked ass. Service has gone out every single day and when it finally came back up my speeds never surpass 30Mbps. After multiple calls it will go back up to 900Mbps and work for a few hours than boom stops working again. Hasn’t worked properly since I go it.

I’m moving soon and the area I’m going to offers Fiber up to 1000Mbps via AT&T @$70 a month so I plan to make the job as the experience so far with what I have has been terry

 

[TheLostSwede]

At that point you aren't really a home user, you're a business user, get a business connection if you want. But for home use, 40Mbps is more than enough.

DOCSIS 3.1 can do 1000/1000 symmetric. It's just not something most home ISPs will give because it isn't necessary for the large majority of home users.

Dude, not everyone lives in the US. There's no such thing as a business connection from my provider.

Based on your logic, no-one needs more than 640Kb of RAM either then..?

What's necessary or not isn't up to you, is it? It comes down to every person's individual needs, wants and willingness to pay, no?

 

[Bill_Bright]

Dude, not everyone lives in the US. There's no such thing as a business connection from my provider.

I find this curious. Don't some folks run businesses out of their homes in Taiwan? For example, have a little computer repair shop, small engine repair, licensed day care, or perhaps a photo studio in their homes? And if so, are you saying their Internet accounts are just normal residential accounts?

It is not just about getting more bandwidth (if needed) for a business account. There are other legal ramifications too. For example, a business account may have a more robust SLA (service level agreement) where contractually, the ISP must ensure connectivity 24/7 and/or guarantee priority repair response - since lost connectivity may mean lost business. Different taxes are involved too.

If running a business out of ones homes are legal in your country, I find it hard to believe business accounts with the service providers aren't available too.

 

[TheLostSwede]

I find this curious. Don't some folks run businesses out of their homes in Taiwan? For example, have a little computer repair shop, small engine repair, licensed day care, or perhaps a photo studio in their homes? And if so, are you saying their Internet accounts are just normal residential accounts?

It is not just about getting more bandwidth (if needed) for a business account. There are other legal ramifications too. For example, a business account may have a more robust SLA (service level agreement) where contractually, the ISP must ensure connectivity 24/7 and/or guarantee priority repair response - since lost connectivity may mean lost business. Different taxes are involved too.

If running a business out of ones homes are legal in your country, I find it hard to believe business accounts with the service providers aren't available too.

For one, there are no data caps here, so that's not an issue.
Secondly, there's only company that I know of that provides business accounts, but there's really no difference in terms of service or hardware. You don't get better support either. Then again, it doesn't seem to cost more either, so I'm not sure how they differentiate business and personal service.

Unless you're an enterprise, there's no such thing as an SLA here and no guaranteed service or response time. That said, even for consumers, my ISP tends to fix issues within hours, even when there's a typhoon and the service goes out due to flooding or something else, it's been back up 3-4h later.

Perfectly legal, a bit of hassle, but no, no special internet service for businesses from most of the service providers.

As per the bottom of this page, SMBs are recommended to use FTTB (which is fibre to the curb really, then VDSL2/2+) or ADSL...
Maintenance service is only working hours during weekdays on those services.

HiNet

 

[newtekie1]

Where I live I was charged $50 extra for unlimited data when using my own modem.

Yeah, but you can rent their modem for $15 and get unlimited, so there's no point in owning your own if you want unlimited.

But to everyone, I plan to switch services as every since I upgraded my speeds to 1000Mbps my internet has sucked ass. Service has gone out every single day and when it finally came back up my speeds never surpass 30Mbps. After multiple calls it will go back up to 900Mbps and work for a few hours than boom stops working again. Hasn’t worked properly since I go it.

You probably need a booster in the box on the outside of you home. They had to put one in when I upgraded to 1Gbps.

Dude, not everyone lives in the US. There's no such thing as a business connection from my provider.

Based on your logic, no-one needs more than 640Kb of RAM either then..?

What's necessary or not isn't up to you, is it? It comes down to every person's individual needs, wants and willingness to pay, no?

Then just change providers. Isn't that what you told all of us with crappy uploads to do? If you seem to think it is so simple for us, then it must be super simple for you. So just switch to a provider that does offer business connections.

 

[TheLostSwede]

Then just change providers. Isn't that what you told all of us with crappy uploads to do? If you seem to think it is so simple for us, then it must be super simple for you. So just switch to a provider that does offer business connections.

You clearly didn't bother reading my post above about business connections not being a thing here...

 

[Durvelle27]

Yeah, but you can rent their modem for $15 and get unlimited, so there's no point in owning your own if you want unlimited.



You probably need a booster in the box on the outside of you home. They had to put one in when I upgraded to 1Gbps.




Then just change providers. Isn't that what you told all of us with crappy uploads to do? If you seem to think it is so simple for us, then it must be super simple for you. So just switch to a provider that does offer business connections.

That’s exactly why I have their modem now

And the tech state that the node was upgraded in my area and that they aren’t allowed to use amps or boosters anymore.



I have another question

So like mentioned I’m moving soon and the new place has a work shop that I will use as my office but it’s not near the central part of the house; I was thinking of running a Cat5e Ethernet cable to the shop than use my TP-Link as a AP and wire all the equipment In the shop to the router with patch cables for internet. Would the speeds be affected by the distance or by me using the router as a AP instead of a router.

 

[newtekie1]

And the tech state that the node was upgraded in my area and that they aren’t allowed to use amps or boosters anymore.

They will install one if you complain enough. You could also have a direct cable running from the box on the outside of the house to the modem, no spices if possible and definitely not splitters.

I have another question

So like mentioned I’m moving soon and the new place has a work shop that I will use as my office but it’s not near the central part of the house; I was thinking of running a Cat5e Ethernet cable to the shop than use my TP-Link as a AP and wire all the equipment In the shop to the router with patch cables for internet. Would the speeds be affected by the distance or by me using the router as a AP instead of a router.

If you turn DHCP off on the TP-Link and plug the cable into one of the LAN ports, you shouldn't see any affect on speed. But if you are running a cable anyway, I suggest running CAT6. Not that it will make a difference, but if you're doing the work to run the cable, might as well do it right.

 

[TheLostSwede]

That’s exactly why I have their modem now

And the tech state that the node was upgraded in my area and that they aren’t allowed to use amps or boosters anymore.



I have another question

So like mentioned I’m moving soon and the new place has a work shop that I will use as my office but it’s not near the central part of the house; I was thinking of running a Cat5e Ethernet cable to the shop than use my TP-Link as a AP and wire all the equipment In the shop to the router with patch cables for internet. Would the speeds be affected by the distance or by me using the router as a AP instead of a router.

No, not as long as it's connected via cable to the main network.
All you do is go to the Operation Mode setting of the C7 and set it to be an AP.

 

[John Naylor]

To get a true test .. you have to shut down all other devices on your network.

But frankly, I just noticed that my 200 Mbs service cost has doubled .... and in the 2years we have had it, it's not like it has changesd our lives in any way. Unless offered a significant discount ... we will be dropping back to basic level Mbps service. Not bringing anything to the table except bragging rights.