Extremely high constant packet loss on EU servers
It seems like when they completely moved to Amazon AWS, and then completely switched everyone to Steam Data Relay (from my observations, this was right at the end of Season 28, in late april), their EU-backbone provider stupidly started enabling shaping and cutting UDP traffic on many input lines. Because of this, we're getting severe losses, even though there are no problems with the server, which doesn't accept half of your packets. The local provider, as always, tells you to **bleep** off and responds with the standard line: "I don't route the game traffic." This also strongly resembles poor optimization of the game itself, because when you bring another window to the foreground instead of Apex, the packet loss disappears.
Update/Additional Info: I’ve done some deeper investigation into why the loss disappears when the game is not in focus (Alt-Tabbed). This strongly suggests that the issue is tied to PPS (Packets Per Second) throttling or a conflict with DDoS mitigation filters on the server side (or at the ISP interconnect level).
Here are the facts:
Frequency Dependency: When the game is in the foreground, it pushes a high-frequency UDP stream. This triggers a consistent ~44-50% outgoing loss. When Alt-Tabbed (game reduces polling/update rates), the loss drops to 0%.
Region Specificity: This is ONLY happening on EU nodes. US and Asia nodes work perfectly with 0% loss (tested via long WinMTR sessions).
Network Integrity: Direct connection to the terminal (bypassing router) and using static IP doesn't help. Standard ICMP pings and traceroutes to the same IPs show 0% loss.
It seems like recent security patches aimed at closing exploits (specifically targeting non-standard port usage or UDP amplification vulnerabilities often used by malicious actors in the RU region) might be misidentifying legitimate high-frequency traffic from certain Eastern European ranges as a "threat." If the server-side firewall or a mid-hop scrubber (like Akamai or AWS Shield) is set to drop "bursty" UDP traffic from specific ASN origins, that would explain exactly why this is happening to us while US servers remain unaffected.
