• Got 10 more DDoS protected servers in EU (OrageBrigade)
  
• Marked the DDoS protected servers in the list with (Anti-DDoS). 18 total for now.
  
• Ended the 3 servers in the 133Hz experiment, reverted back to stock

I'd rather downfrag my own site than upvote a nutshiny post.

serveme.tf is free to try, so you could always give it a go. Unfortunately only the BeretBrigade servers have DDoS protection that really works. That machine is hosted in France though, so might not be good for you in Latvia.

NFOservers has decent DDoS protection in the US, I don't know if their EU servers have the same protection, but the NA stuff is good enough for most attacks. They filter the attacks upstream so they stop hitting your server.

Hiperz is still good in my opinion, I really like the ability to rehost your server to a different region/location if you have bad routing for example (this does cost extra).

A dedicated machine from OVH in their GAME range comes with a proper anti-DDoS solution that's tailored for game servers. You're talking 40+ euros/month though, so you might want to look for someone to split the bill with.

It's not so funny if you're the one paying for the servers.

The best player in the world, who made the best team in the world, on his way to win the biggest prizepool TF2 has ever seen, should have wished his opponents a speedy recovery. You know, sportsmanship, or being a normal human with empathy.

I'd just like to say the stream looks great. Very smooth 60FPS, great encoding quality and good observing, thank you <3

http://fakkelbrigade.eu/i/vanillamarvin_varwwwvanilla_2018-01-29_12-44-25.png

???

https://blog.gravatar.com/2010/04/01/unicorn-gravatars-stack-overflow/

Crap, missed that. Can you record your own vid with net_graph 5?

Check if your router has flood or DoS protection. If it does, disable it.

Your WinMTR shows two routers (192.168.1.1 and 192.168.0.1). Not likely to be the cause here, but undesirable. There's at least 1 ping spike to the 192.168.0.1 router, which is odd if it's on the local network.

I've not seen you mention reinstalling TF2 (and starting completely stock first, no fancy configs) and reinstalling Windows, try those, in that order.

I see you've tried using your PC somewhere else and getting the same problem, but still, can you describe your network setup? Is it a true ethernet connection or something like a powerline adapter? Because those are almost as unreliable as wifi for gaming. What router(s) are you using? How are you connected to these routers exactly?
What kind of network adapter do you use in your PC?

Judging from your video, something happens causing you to receive less packets (see "in" packets/sec drop below 66 when it lags), but at the same time you appear to receive more data (see the "in" k/sec). This is counter-intuitive, but could make sense if these packets arrive out of order or are somehow corrupted / unusable.

You tell me, please :)

Machine ran out of disk space, fixed now :)

The max CPU and GPU temps in those screenshots aren't from a game/loaded state. Make sure HWMonitor is on during your game and check the max temperatures after you've experienced those 40FPS drops.

Do you get the same drops in a synthetic benchmark like Unigine Heaven? If you're playing in windowed mode, does switching to fullscreen mode help?

Depends on the results of the test. Personally, I hope people won't find 133 tick a big improvement since it will increase the CPU load on the servers and I have serious doubts that my two 12-core machines can run 133 tick all the time for 6v6, let alone 9v9.

It's really easy to view the server tickrate (net_graph 5 in the client), so it would be hard to do a blind test.

Check if the modem has a firewall log, some go into a blocking mode if they detect portscans (even outgoing).

Whenever you open the TF2 server browser or Mumble server list, your PC will send out a bunch of UDP packets trying to get info on the servers, some stupid devices with built-in firewalls consider this port scanning and will block connections for a while.

The point is reduced transmission times and fitting all packets within sv_maxrate, it's a trade-off. The time you lose (<0.02ms) to compress a packet is gained by having to spend less time transmitting it (because it's now smaller).

I'd stick with the defaults (both client and server) over lists of TF2 client and server cvars with no data/benchmarks.