vnb wrote: ↑26 Jul 2026, 08:18
xVenom wrote: ↑24 Jul 2026, 00:08
vnb wrote: ↑20 Jul 2026, 17:24
joseph_from_pilsen wrote: ↑20 Jul 2026, 15:30
Active powered fiber line, no PON.
I meant classic old style when people were connected trough SFP modules in smart switches directly to opposite SFP module at providers side again in some local smart multislot SFP switch (and very often combined with 1gbps local ethernet as a last mile and handover interface from smart local switch somewhere in basement). It was a little bit less reliable as it was prone to power outages, but for gaming this connection was way better than PON which is from my point of view a slight downgrade (more reliable but there is always a milisecond or two added in term of ping and jitter).
oh yeah, i see what you mean now. in Belgium this was never available to "normal" customers but i know that some hospitals, banks, airports, etc do have this technology available for them.
At work we called this (P2P - Point to Point) which is essentially a dedicated fiber directly between the provider's central office and your location. Are you sure that you used at some point this kind of technology? Is it extremely restricted to very specific uses here, nobody can access it here except if you are worth multi million dollars...
Btw i agree that PON is definitely not the best fiber optic system. And i wouldn’t call it more reliable either, there are tons of issues. But it is the one that can fill up ISP's pockets the fastest; you can literally split A SINGLE Pon link to 32 to 64 customers -- that is 2,5 Gbps down 1,25Gbps UP shared across 64 houses.... so no wonder why all the money hungry ISPs corporates moved to use it...
You've got the general concept right, but the issue is happening further down the line if at all, specifically after the junction point (where the all connections connects to olt).So in general PON or XGS-PON is really good at this time.
I wouldn’t call GPON or XGSPON "good" after working for years on the field. How can you call this fiber technology good when it is shared with 32 or 64 other customers? They moved from Docsis because it was said "this is a shared tech between customers; bad bad bad" then ISP chose to implement Fiber PON, which is what? You guessed it a shared fiber technology LMAO.
Let me give you an example why IMO shared PON is bad : there is a technology called called Dynamic Bandwidth Allocation (DBA). Essentially it is an algorithm in the OLT (ISP side) which grant upstream slots based on reported queue status.
Simply said: the more data you upload, the more the OLT grant your ONT based slots time.
It is like if a teacher will give more attention to the kid that talks a lot in class and less attention to the quiet one (the ONT that does not upload a lot).
So essentially the more you abuse the upload, the more the OLT gives you "priority" meaning upstream slots...
Let's get two things very clear. First, a 'behind the server' feeling or high buffering is almost always caused by downstream issues, not upstream. Hit registration depends on both: upstream sends your action, and downstream brings the server's confirmation back so your client can display the kill mark.
As for OLT time slots: I've been down that road. There were days I'd use hping3 to send over 3,000 pps to my ISP's first hop IP just to lower my standard deviation (sending it tagged as CS0 while keeping my game packets tagged as EF with a shaper at my WAN so game packets wouldn't get queued). The idea was to claim more OLT upstream time slots. But did it actually make a difference in-game? None, unless your upstream is already very jittery. Any jitter value below half the server's tick rate is considered very good.
Now regarding shared bandwidth: GPON speeds are 2.4 Gbps download and 1.25 Gbps upload. Assuming a max plan speed of 1 Gbps down and 500 Mbps up (depending on the provider), it takes only two and a half users maxing out their connections at the same time to fully saturate the node. Theoretically, it's possible, but practically, upload saturation is extremely rare. In an extreme case where two or three users download a 50GB game simultaneously, the saturation only lasts about 6 minutes (assuming their hardware can even write at those speeds). In a general case, a 4K Netflix stream takes 25 Mbps max, meaning it would take 50 simultaneous 4K streams to max out GPON download. So why do some connections feel worse during peak hours? It's almost always on the download side, but it's typically driven by routing and peering issues rather than local node congestion.
XGS-PON (10 Gbps) is a completely different story in practice, it's almost impossible to saturate. Almost no consumer devices except PCs or servers can even pull 2 Gbps, and even if they do, it's only for a minute or two. You can download 100GB in under 4 minutes, and nobody has the storage setup to download continuously at that rate indefinitely.
While I'm not ruling out GPON download saturation in extreme edge cases, upload saturation is exceptionally rare. The bottleneck is almost always after the OLT, in ISP routing and peering. I used to think GPON and XGS-PON were bad for gaming, but they really aren't for today's demands. GPON might show its age sooner, but XGS-PON will easily be great for at least the next 10 years.
You can take a look at this study on gpon saturation from a isp provider moderator. use translation, though it doesnt fit every variable but it gives your good clarity on this context.
https://lafibre.info/gpon/le-gpon-ne-sa ... -la-preuve
Also Remeber every player uses the same pon tech.
