PyCCo_TyPuCTo

Oh I notice it has Killer software, is that installed?  That often lives up to its name and kills the speed.

AH there it is, i somehow haven't noticed this running for the 3 years ive had this setup. probably because its not "really" a problem lmfao. Removed the service software and kept the drivers Ethernet speed now matches wifi with 2ms better latency. thanks!!! i would have never thought to look for it because its not in my startup programs, doesn't pop up every and not in my system tray. I literally had to go through and search for the program itself its extremely well hidden.

To answer your question CRCs increment when a corrupted packet is detected (and dropped). A ton of different things can corrupt a packet.
You actually have really good SNR and attenuation values, unless these numbers fluctuate you should be good there. Are your CRCs intermittent or is it a slow trickle? If they come in in bunches, are you saturating your link during those time frames? Saturation will cause CRCs as well.
Have you tried complaining to the ISP and have them run their checks remotely? There could be a bridge tap somewhere on the path between the CO and your home.
They can also try and play with the DSL parameters to try and improve things, but that's granted they see something not quite right on their end, and actually care about doing something. I'd imagine residential DSL is not going to be looked at by higher tier technicians.

If you want new,
 
TP-Link TL-SX3008F with 8 10g SFP+ ports is $232 : https://www.amazon.com/TP-Link-TL-SX3008F-Enterprise-Integrated-Protection/dp/B0973T7BGL/
The 16 port version is $500.
 
Microtik has this $262  8 x 10g SFP+  and 1 x 1 gbps switch (CRS309-1G-8S+IN) : https://www.amazon.com/MikroTik-Desktop-Gigabit-Ethernet-CRS309-1G-8S/dp/B07NFXN4SS/
 
 
there's a Mokerlink 8 port SFP+ switch for $215 on amazon but it's not rackable : https://www.amazon.com/MokerLink-Support-Bandwidth-Unmanaged-Ethernet/dp/B09W24RZDC/
 
QNAP has some reasonably priced models if you look hard
for example
369$ for  6 x 2.5g rj45 + 4 sfp+ 10g : https://www.amazon.com/QNAP-QSW-M2106-4S-US-10-Port-Managed-Deployment/dp/B0BK2M7X7M/
 
Netgear has a $500 switch with 4 x 1G/2.5G RJ45 and 4 x 1G/2.5G/5G/10G RJ45 and 2 x 10G SFP+ ports : https://www.amazon.com/NETGEAR-10-Port-Multi-Gigabit-Ethernet-MS510TXM/dp/B08PPYTB1V/
 
 
 
For used stuff, freshly pulled from datacenters or new old stock or sometimes refurbished,
 
 If you're willing to deal with a discontinued model, you can get a Cisco 6120XP  (N10-6100) switch for $210 , which gives you 20 SFP+ ports: https://unixsurplus.com/cisco-N10-S6100-20x-10gb-sfp-switch/
It can be flashed to a Nexus 5100 supposedly, see review of 6120XP and discussion on reddit thread here : 
The same company I linked to above, sells  Quanta T5032-LY6 switches with 32 x 40g QSFP ports for $419 + $25 shipping : https://unixsurplus.com/quanta-t5032-ly6-switch/
 
DAC cables that break out a 40g port into 4 x 10g ports start from around 30-40$ (for the 1m/3ft lengths).
 
 
You can also look on eBay for Arista or Juniper 10g switches, you'll find plenty.

Just stick with 1Gb, save your money.
 
If you want to be able to use 5Gbps at each host you need Cat6 wiring throughout the property, as well as 10Gb switch.
Depending on number of ports and PoE requirements, managed or unmanaged, etc... these puppies can get pricy

Just stick with 1Gb, save your money.
 
If you want to be able to use 5Gbps at each host you need Cat6 wiring throughout the property, as well as 10Gb switch.
Depending on number of ports and PoE requirements, managed or unmanaged, etc... these puppies can get pricy

Just stick with 1Gb, save your money.
 
If you want to be able to use 5Gbps at each host you need Cat6 wiring throughout the property, as well as 10Gb switch.
Depending on number of ports and PoE requirements, managed or unmanaged, etc... these puppies can get pricy

If the ISP does CGNAT then port forwarding does jack shit. Because you're still under another layer of NAT. 

Cellular connections are typically behind CGNAT and are not able to accept any incoming connections. 

Comcast has ran trials in select markets and was able to do Gigabit upload. Currently they are deploying mid splits and lower tiers are getting 100 Mbps upload while upper plans are getting like 500 Mbps upload. So thats no longer going to hold true. 

Why is everyone so sure OP is behind CGNAT?! It sounds like he should have a public IP on his AT&T router, or since it's bridged - on his UDM Pro...
 
@Sekigema 
IF by "passthrough" you mean that the AT&T router is in full bridge mode, then your UDM Pro should be getting public IP. In that case no need to set up any port forwarding on the AT&T router, but just on the UDM Pro.
In fact, since it's in passthrough, disable everything within it, you want it to be as dumb as possible and just forward everything to UDM Pro
https://community.ui.com/questions/BGW320-500-Bridge-Mode-and-or-IP-Passthrough-Question/99786f13-1f76-46dd-9801-7102fd1d44d7
 
The above plus this guide should work:
https://www.hostinger.com/tutorials/how-to-port-forward-a-minecraft-server 
 
Even though your IP from AT&T is not static, it should still be unchanged for extended periods of time, and shouldn't cause you issues.

I would download the windows store app for Speedtest.net and call it accurate enough.
 
https://apps.microsoft.com/store/detail/speedtest-by-ookla/9NBLGGH4Z1JC?hl=en-us&gl=us&rtc=1

There's no way you could pull 100' through tight conduit without damaging the individual wires, especially if there are any turns. If you have the wire to spare and you want to try, go for it. But I give it a 99.9% chance you'll be wasting your time.
 
I also just realized, you really shouldn't run your ethernet parallel to (and in the same conduit as) AC lines, if that what's you meant by "power cable for the garage."

That should work.

I have a similar setup at one of the businesses provide IT services for. One thing to note is if you're on Bell Aliant (Atlantic Canada) this doesn't apply. 
 
First there's two pieces of information you're missing which is why you just can't connect your own router to the Nokia ONT.
1. You need your PPPoE credentials. These are different from your Bell Hub login credentials. Assuming you're on a business account it would be best to call Bell and have them provide these. The user ID will be XXXXXXXX@bellnet.ca or possibly b1XXXXXX. 
 
2. From the Nokia ONT Bell uses VLAN 35.
 
From there you just need to configure your router WAN interface to use VLAN 35 as well as PPPoE with the credentials provided by Bell. 
You didn't mention what router you intend to use or what the internet speed is. You might need some tweaking if you're aiming to hit 1Gbit symmetrical for example. PPPoE performance can be hit or miss depending on the router brand/model.

Tested: Latvia - NewYork city
I would say 10/10 ... and i pay 18 eiro (~ 20 dolars ) for this and TV ... so i guess its not bad at all

I've been having this too recently, seemed to happen after a Windows Update, although some people say its the card overheating and can be helped by adding a fan blowing on the card.
 
I've had to revert that machine back Gigabit for now as I couldn't download anything on Steam.
 
Its telling as I have the same card on a Linux machine and it has zero problems, I may have to try swapping them.

Yeah, they work great in my experience. Certainly better than powerline if you have coax available to you. Make sure any splitters, amps, or couplers are MoCA rated (i.e. >1GHz).
 
EDIT: if this is going to be the only link in your MoCA system, I would grab the MoCA 2.0 starter set (2 pack) which is currently 46% off (making it cheaper than one 2.5 adapter). MoCA 2.5 only becomes useful when you have >2 endpoints (or actually want 2.5G speed, but then you need the 2.5G ethernet adapter version).

Don't you LOVE ISPs ...

This is due to tcp window size getting bigger as the download progresses. the packets in the beginning will be smaller, and if there are no errors (re-transmissions) the file transfer protocol will increase the next packet size, while overhead remains the same. This gives you a better ratio of user data to overhead, resulting in higher download speed.
on fiber, errors are rare, but if an error does happen and packet needs to be re-transmitted, then the window will get smaller and you notice a slight drop in speed before it goes up again. Happens a lot on DSL lines. Fiber is usually pretty good, errors are rare.

This is due to tcp window size getting bigger as the download progresses. the packets in the beginning will be smaller, and if there are no errors (re-transmissions) the file transfer protocol will increase the next packet size, while overhead remains the same. This gives you a better ratio of user data to overhead, resulting in higher download speed.
on fiber, errors are rare, but if an error does happen and packet needs to be re-transmitted, then the window will get smaller and you notice a slight drop in speed before it goes up again. Happens a lot on DSL lines. Fiber is usually pretty good, errors are rare.

Yup. The proper way around it is to use an FQDN that you resolve yourself when inside your network to point to a local address, but resolves to your external IP when out in the world.
You can do a "hacky" workaround if your firewall gives you that level of control, where instead of allowing packets destined for [public ip]:[RC unit port] from inside your network to reach it, you redirect them back inside to the correct local address, but it's a pain to maintain if you don't have a static IP.
You've already done that, or it wouldn't work when you are off the local wifi, probably via some UPNP mechanism.
 
TBPH, If I was looking to control hardware remotely, I'd be doing it via a key only SSH session, that way whilst the "door to everyone" is visible, it's clearly locked shut.

The connection from your home to the other end is usually shared with other customers.   Example, if its GPON based then you are sharing 2.4Gbit down, 1.2Gbit up with potentially 32-64 other people who are connected to the same fibre.  What you pay for is how much of that capacity you have access to.  Usually you will be able to hit (as not everyone will be using it that much) your cap, but at peak hours the odds of the connection reaching capacity is increased.
 
Even if its dedicated fibre for you alone, somewhere up from that it will be shared over something like a 10Gbit fibre link which is ALWAYS smaller than the combined speed of everyone connecting to it.  Its simply not feasible to have enough bandwidth for everyone to be maxed out when the law of averages means 99% of the time very little of that bandwidth would be in use.

The entire Internet works this way but the amount of bandwidth goes up exponentially the further up the chain so contention gets less and less likely.

Contention is when you have say four customers on Gigabit all trying to pull full speed from that 2.4Gbit shared connection, we'd call that a 2:1 contention ratio as there is only half the bandwidth available than would be needed to avoid a bottleneck if everyone is downloading as fast as they can.  The actual contention ratio is likely a lot higher but they're not going to tell you what it is and a good telco/ISP should monitor it to ensure most of the time its never maxed out.

By further up the chain I merely meant where the various different parts of the network connect to each other.  In the above example, your 2.4Gbit PON might connect to your ISP over a 10Gbit fibre link shared with (just an example) 10 other 2.4Gbit PONs.  So you can be bottlenecked at various points, but as its unlikely every customer HAS Gigabit or that many customers will actually be pushing full speed at the same time, you'll generally not see it.
 
I've seen some articles claim contention is not an issue on fibre, but this is not true at all.  Some ISPs will sell 10Gbit fibre over a 10Gbit PON, so one single user could (in theory but unlikely) max it out.  Its all based on calculations by the ISP of what the "average" use will be so that the majority of the time every user will achieve full speed.

A dedicated 10Gbit link to your ISP is insanely expensive, this is the kind of thing data centres or big businesses would use if they need guaranteed capacity.  You have to remember the speed OUT of your ISP will be far less than every customer combined could use, so with a dedicated line you're paying a premium for them to ensure those links are big enough to guarantee your speed also.

If modem was replaced and there is nothing wrong with the cable pair, but the issue persists, then keep pushing the ISP because they often don't find a problem the first time around but later if you keep pushing they fix it.