Charging for 200mbps while delivering 60mbps
Charging for 200mbps while delivering 60mbps
He mentioned his second test came from mobile fast.com, which uses Netflix's speedtest servers and isn't necessarily hosted on his ISP's network. This helps verify his connection speed. I checked older ISPs with weaker hardware and heavy NAT settings—they often struggled to reach the 100Mbps limit. No, it’s not due to that. Generally, big ISPs serve many users with varied speeds. The guaranteed minimum speed (CIR) is much lower than what his plan supports. I’m not suggesting this is the sole reason for his slow performance; I just want people not to expect constant 200Mbit even if they can reach 1Gbit most of the time.
It was my mistake to respond so hastily; I’ll keep this in mind to take time and read more before sharing.
In the bigger picture, wired connections don't really matter much. What matters most is making sure his WAN gets enough throughput. Most ISPs offer direct CDN peering, which gives full performance. The original issue was getting 200 Mbps until he upgraded to a new motherboard or CPU. The router itself doesn’t seem to be the problem. CIR isn’t a fixed minimum—it’s the highest possible. With consumer connections, speeds on shared networks like GPON or cable aren’t always guaranteed. The actual limit is set by the port and chassis uplink, and your bandwidth is controlled at the rate you pay for. You can’t always expect 100% speed, but there are no fixed guarantees either.
It seems like there were some problems initially. Everything functioned until the hardware was changed. For an accurate speed test, you need just one device connected—no other devices on Wi-Fi or Ethernet. Many people notice complaints about slow speeds, but it’s often because their computers are running P2P software or malware.
It varies based on your ISP and country. In the UK, Openreach is enforcing guaranteed minimums for FTTP connections, which likely influences their contention levels on the PON. Their DSL services also have limits—dropping below certain sync rates lets you switch packages even during the contracted term.
These aren't CIR either. With PON the limit is max 32 per PON due to loss per splice. Then ratio out is capped per customer. That doesn't ensure anything beyond preventing over-subscription. The second case simply lets you exit early if the copper quality is poor. CIR is a queuing technique, neither of these fit that description.
I thought the PON would allocate a baseline time slice for each customer to ensure consistent performance. It seems there might be contention factors involved, which could allow a high-speed user to overwhelm others during busy periods.