Internet provider options what would you do?

Anything over a Gbps is mostly marketing. There are very few people that could even utilize that amount of bandwidth without expensive enterprise class networking hardware and wiring in your house. Most inexpensive home networking equipment is Gbps with 2.5Gbps getting a little more affordable. Most people's perception of bad internet comes down to poor equipment or configuration problems.
I just switched to 1GB service from 80Mbps DSL only because it was cheaper to switch companies. I worked from home a lot as an sw engineer, had DirectStream TV on multiple Rokus, and 3 kids that gamed and did whatever teenage things without a problem on 80Mbps. Kids have moved out, but I can't really tell the difference with the 1GB unless I need to download a large file. The only complaint when the kids were at home was when they downloaded a large PS5 game. It took an hour or 2 instead of 10-15 minutes.
I couldn't agree more!! (if you seen my others posts in here for years) That is why I posted for $49 - 350/350

Im so annoyed as my new home and internet provider is forcing me from 300/300 (that was already 57.95) to 500/500 at a higher cost of 67.95). I have no options and we do not need the speed. It's all profit for the companies. Worse is that our company is a CO-OP but the public just pays and I bet 90% or more automatically choose the higher speed because they think the other offerings are to slow. Actually I KNOW they do, we are a large community and I see it in their Facebook posts.

Another game is some providers give a cheap modem/router with lower cost plans, when people have issues they are sold on "upgrading" to a higher plan, even though it is just the terrible router OR placement.

I have worked on/in security many homeowner and business wifi networks. Most are a disaster. Though has gotten better over the last decade.
Most all homes can easily get by on 200/200
ITs ALWAYS the network in the home if someone has a problem with those speeds at the modem

I bet 90% of homeowners never even bother to check their speed and see if they are getting what they pay for. I do and when I dont I call my provider. If unhappy I file a complaint with the FCC and guess what? IN a matter of a day or two you will have supervisors calling your home and asking when they can come over to fix the issue. They will be falling all over you, even once they are done will follow up with calls making sure you are happy.

I filed on my last home ... they worked on my street for hours. I have done the same with this new home, my upload speeds were not symmetrical with the download. 500/500 I called the company, got an annoying person who answered the phone, she gave me a rough time, was nasty. I just said ok, Im filing a complaint with the FCC and I did, it only takes minutes.

Within 48 hours or less I got a call from "Higher ups" along with their cell phone numbers and asked if they could send someone right over that afternoon! Huh imagine that! Guy spent hours here trying. literally, on the phone with his company at the tech end, with his laptop, upload speeds were not what I was paying for, I wasnt going to let this go since they were raising my rates and speeds by force.
Anyway, long story short, not kidding, hours, he finally said let me try changing out the 2 year old modem and if that does not work he has a high powered one but its much larger and will not fit in my service panel so he tried the same model that was there and it worked.

If anyone in the same situation here you go - https://consumercomplaints.fcc.gov/...nternet-Form-Descriptions-of-Complaint-Issues

Internet providers DO NOT like to get complaints, congress reviews complaints they will fall all over you trying to make sure you are good. But just remember first to make sure you are getting your speed readings from your modem and its not your router or set up... but then again, if unsure you can still call your provider and if not happy, then the link above will take care of that.

@tmorris1
(just put this here in case you read it before I was done with my edits)
 
Last edited:
This is a good tool. I would also suggest being very careful with trying to fix the findings because it's quite a complex topic.
Agree. I think if the tool gives you an F or a D, you should be able to easily improve things, possibly just by getting a more modern/better router. But if you're already a B, trying to improve to an A, or if you're an A trying to improve to A+ may be a lot harder and possibly not worth the hassle.
 
When you throw in speed differences, like a 2.5g internet (link speed - not committed bandwidth) and a 1g link to a desktop, you need added buffers to avoid drops from the speed difference. Things coming in at 2.5g rates can't leave as fast on 1g, so you have to buffer some or you'll get packet loss that hurts your overall throughput.
That's a very complex topic. A 2.5Gb/s internet with a switch interface of 1Gb/s is a differential speed, but it's highly improbable for a 1Mb/s UDP gaming stream to overrun a L2 buffer on an unloaded 1Gb/s switch port. It's not impossible, but high improbable. If there is also concurrent bursty download traffic, then yes the 1Gb/s switch port may have momentary congestion, but that's not caused by the gaming traffic. If buffers are increased on the 1Gb/s Ethernet switch port, the Ethernet frames will spend more time sitting in the buffer before being dequeued, which adds latency. It's better to use class of service at layer 3 to prioritize the latency sensitive traffic and demote the dequeing probability of the bulk traffic. Punish the bulk traffic that is not latency sensitive so that it doesn't step on the latency sensitive traffic. This method has much better results than increasing Ethernet buffer size.
 
I don't disagree with any of that, I'd only add the solution is sometimes not buffer tuning or QOS action (perhaps for elephant flows), it's often both.
 
I don't disagree with any of that, I'd only add the solution is sometimes not buffer tuning or QOS action (perhaps for elephant flows), it's often both.
Some older Ethernet switches had tiny default buffers and were terrible about dropping frames. It's been many years since I had to increase buffer size on Ethernet switches, instead properly configured L3 CoS is the only remedy needed on modern enterprise hardware. "Properly" is the operative term. L3 Cos is also highly highly complex and I know very few people that can get it right. In my opinion even Cisco's book on CoS is mostly wrong. Cisco's recommendations work in a lab environment, but fall way short in a production enterprise environment.

Some network administrators work to eliminate all packet/frame drops and that is just wrong. There are times you WANT frames/packets to be dropped. Dropping packets is a tool to slow TCP flows. When you are in the middle of the flow, the only real mechanism you have to slow a TCP stream is to drop packets.

I understand we aren't talking about enterprise hardware and environment and not all concepts are the same, but they are similar. One of the major differences is consumer hardware is lacking in so many aspects.
 
Last edited:
What do you mean by L3 CoS?

CoS is in the VLAN tag (802.1p ) and DSCP is in the IP header (L3).
 
What do you mean by L3 CoS?

CoS is in the VLAN tag (802.1p ) and DSCP is in the IP header (L3).
CoS, QOS whatever you want to call it. The terms are used interchangeably in by network admins. The VLAN tag is for identifying allowed ports, not identifying layer three priority. An entire VLAN can be prioritized on an Ethernet switch, but that is such a broad paint brush that it's not that useful. The Type of Service (TOS) field is in the IP header (L3) and is used to carry the ToS, DSCP, and ECN values for the L3 packet.
 
The terms are used interchangeably in by network admins.

That's not been my experience.

The VLAN tag is for identifying allowed ports, not identifying layer three priority.

Well, it's for Layer 2 priority. In the 802.1P tag you have the VLAN ID and you have 3 bits reserved for the CoS / Priority. This is what's used in DSCP to COS maps and the reverse, especially when you're talking about trunk links (like to phones).
 
That's not been my experience.
You use the term correctly, I have talked to hundreds of network admins over the years and probably 75% don't use the QoS/CoS terms correctly in context of L2 or L3.

Well, it's for Layer 2 priority. In the 802.1P tag you have the VLAN ID and you have 3 bits reserved for the CoS / Priority. This is what's used in DSCP to COS maps and the reverse, especially when you're talking about trunk links (like to phones).
Oops, I didn't see p in "802.1p" The "p" is the Ethernet QoS standard marking, similar to the L3 IP TOS field.
 
Oops, I didn't see p in "802.1p" The "p" is the Ethernet QoS standard marking, similar to the L3 IP TOS field.
Because I worked in core with very fast EtherBundles (now > 16Tb/s) for so long, you can probably tell that I don't really care about L2 QoS.
:)

Yes, even bundles running at those speeds utilize weighted deficit round robin for L3 CoS. There is no such things as no congestion.
 
There is no such things as no congestion.

Exactly. At any given instant, there's no such thing as 50% interface utilization. It's either 100% or 0%. If it's 100%, because it's actively transmitting, the next packet/frame has to queue up.
 
Exactly. At any given instant, there's no such thing as 50% interface utilization. It's either 100% or 0%. If it's 100%, because it's actively transmitting, the next packet/frame has to queue up.
Clearly you know what you are talking about. Trying to get some network admins to understand that caused me hair loss. "But the 5 minute utilization is only 35%, how could there be congestion". UGH.

Do you remember FRF12? Insertion of 1,500 byte packets on 56k and T1 lines was so slow that you had to fragment every packet > a threshold to try to reduce the jitter caused by a nasty large packets getting serialized. The line was 100% utilized the entire time that large packet (and any packet) is getting put on the line. At 100Mb/s and higher, we don't even consider insertion delay, but in the good old TDM days, insertion delay was a big deal.
 
Last edited:
  • Haha
Reactions: Pew
I had to jog my memory on FRF.12 .... LFI! Yes, that I do recall.

I didn't get my SmartBits 600 until a bit later, but that would have been so useful in those times to quantify improvements.
 
That's a very complex topic. A 2.5Gb/s internet with a switch interface of 1Gb/s is a differential speed, but it's highly improbable for a 1Mb/s UDP gaming stream to overrun a L2 buffer on an unloaded 1Gb/s switch port. It's not impossible, but high improbable. If there is also concurrent bursty download traffic, then yes the 1Gb/s switch port may have momentary congestion, but that's not caused by the gaming traffic. If buffers are increased on the 1Gb/s Ethernet switch port, the Ethernet frames will spend more time sitting in the buffer before being dequeued, which adds latency. It's better to use class of service at layer 3 to prioritize the latency sensitive traffic and demote the dequeing probability of the bulk traffic. Punish the bulk traffic that is not latency sensitive so that it doesn't step on the latency sensitive traffic. This method has much better results than increasing Ethernet buffer size.
The problem is worse with asymmetric speeds like 500 download and 50 upload. Even though you aren't uploading a lot of data every download packet requires an acknowledgment back to say it was received. When things get congested everything slows down. If you have a decent router you can usually fix this by enabling Quality of Service (QOS) and enter your upload and download speeds
 
If you have a decent router you can usually fix this by enabling Quality of Service (QOS) and enter your upload and download speeds
Yup, I can turn on Smart Queues (Unifi's version of QoS) on my UDM SE, and it'll get me to A+ on bufferbloat test, but at the expense of lowering max down/up speeds, plus it's more taxing on router's CPU. Not really worth it in my case.
 
The problem is worse with asymmetric speeds like 500 download and 50 upload. Even though you aren't uploading a lot of data every download packet requires an acknowledgment back to say it was received. When things get congested everything slows down. If you have a decent router you can usually fix this by enabling Quality of Service (QOS) and enter your upload and download speeds

Ugh......Rip-my-hair-out issues with s2s VPNs and CAD programs.
 
Yup, I can turn on Smart Queues (Unifi's version of QoS) on my UDM SE, and it'll get me to A+ on bufferbloat test, but at the expense of lowering max down/up speeds, plus it's more taxing on router's CPU. Not really worth it in my case.
I have a Cloud Gateway Max and mine will keep up with Smart Queues turned on at 1Gbps but I have it turned off as well. I don't game and the only place I can tell is on the bufferbloat test website. Ubiquity says not to use Smart Queues over 300Mbps but mine keeps up so newer hardware is more capable and that is old guidance I suspect
 
Last edited:
I have a Cloud Gateway Max and mine will keep up with Smart Queues turned on at 1Gbps but I have it turned off as well. I don't game and the only place I can tell is on the bufferbloat test website. Ubiquity says not to use Smart Queues over 300Mbps but mine keeps on so newer hardware is more capable and that is old guidance mane.
Yeah, I'm on a 300 down / 100 up plan, so technically I could run Smart Queues, but my bufferbloat is already at A. Getting it up to A+ doesn't seem necessary in my case. I think using the Policy Engine/Traffic Rules to prioritize specific latency-sensitive apps is probably a better way to go.
 
Cost - you have great prices and I see competition which is T-Mobile and ATT.
Where I live, we have no option to go lower than 500/500 and the cost is $68 a month/

I did have 300/300 for $58 but they did away with it and forced us to the higher price. We have NO other options and they know it.
Even worse is this is our local co-op here in NC
At our previous house in SC we had our electric utility co-op offering fiber and it was $49 for 500/500

Typical families dont need more than 300/300 and the providers know it so they charge more for speed you dont need = Profits
Definitely a logical reason. It's baffling that the provider would charge you $68 for 500/500.... but at least the cellular service works well enough.

We ended up going with 500mbps from Frontier at $29.99/mo... it's hilarious to see _actual_ usage over time. Our 95th %ile works out to 21Mbps, even with loads of streaming, teleconferencing, etc, etc... and despite the sales attempts of the ISP, this works wonderfully for the nearly 100 devices we have in the house.

1772766487144.webp
 
Definitely a logical reason. It's baffling that the provider would charge you $68 for 500/500.... but at least the cellular service works well enough.

We ended up going with 500mbps from Frontier at $29.99/mo... it's hilarious to see _actual_ usage over time. Our 95th %ile works out to 21Mbps, even with loads of streaming, teleconferencing, etc, etc... and despite the sales attempts of the ISP, this works wonderfully for the nearly 100 devices we have in the house.

View attachment 327083
Oh, I can see how my post cold be mis-read. We have Fiber 500/500 @ $68 a month.

Some people are lucky to have competition from the cell service providers which gives them options.

We have no options (UPDATE we may) just maybe a mile to a mile and a half (at most) to far from the cell towers for T-Mobile, I think AT&T is on that tower too as I switched from T-Mobile on my cell phone to ATT, my wife is on Verizon, we ditched T-Mobile a year or more ago because if their pricing games ... anyway When I drove my truck to a nearby tower back then about 2.5 miles or so away a year or so ago. I pulled 700 Mbps on my cell phone. Pretty consistent from that tower on my at the time T-Mobile Cell phone. But back at my home I can only get at the time around 40 mbps/

Ok, so that was then, I just tested and I am shocked at my ATT iPhone downloads, still not great however much better than T-Mobile ever was and hands down far, far, superior to my wife's Verizon service!
I just tested because of your post. On my ATT iPhone service and I am getting consistent 65 to 80 at my house (one test just pulled 100) Mbps however upload is only 6


Ok, up next I ran my wife's Verizon cell service on her iPhone and it was a horrible 7 to 10 Mbps and 0.1 to 3.0 upload *LOL*
Im glad for your response. I used to test my cell phone speeds all the time but gave up, something happened with ATT being so far superior to Verizon and a lessor degree T-Mobile at our house. Now that my wife just saw her slow test speeds I have to change her over from Verizon to AT&T:ROFLMAO:

Cell phone speed at our house does not matter much since they are on Wi-Fi when home (except when testing) Ive always been a ATT fan here on the east coast and well, I guess once again its confirmed superior for our area.

IN case I confused you with my long rant, our home internet is the Fiber mentioned. But testing cell phone speeds MIGHT mean in the future we have options to go to lower coast cell internet service if prices keep going up on Fiber.

I cant believe you are paying $29.99 is that a one or two year special before they slam you with price increases? Wow that is cheap
For the record, I too, know full well, as I posted many times, people are pushed into higher speeds that they cant use and its all profit/revenue for the providers. Which is why they cancelled my 300/300 service and pushed me into 500/500 service.
 
Last edited:
Back
Top Bottom