The problem is that the existing ECN is too blunt. It’s a situation that AccECN looks to fix. “ECN was originally specified for TCP in such a way that only one feedback signal can be transmitted per Round-Trip Time,” the IETF draft specification for AccECN states.
For basic congestion control that was enough, but modern high-speed protocols need to know how much congestion is occurring, not just whether it happened. AccECN goes much further.
Classic ECN only tells the sender that congestion happened. AccECN tells it exactly how much. That distinction lets senders fine-tune their response instead of just slowing down at the first sign of trouble.
UDP benefits from a timing boost
TCP traffic isn’t the only networking traffic that is getting a boost.
The Linux 7.0 kernel includes a notable optimization to the network stack’s timekeeping mechanism, specifically addressing performance bottlenecks in high-speed UDP traffic.
The kernel now reduces the overhead associated with function calls on critical hot paths. This change is particularly significant because compilers often fail to automate this optimization across the boundary between the core kernel and network drivers compiled as modules.
