Telecom giant Vodafone and strategic supplier Nokia have today revealed that they conducted their first real-world trial of “lag-busting” L4S (Low Latency, Low Loss, and Scalable Throughput) technology on a “live home broadband fibre [FTTP] connection“. The trial was able to significant reduce the delay (latency) time during gaming and videoconferencing by 94%.
For context. Latency is a measure of the time that it takes for a packet of data to travel from your computer/device to a remote server and then back again (ping). The delay is measured in milliseconds (e.g. 1000ms = 1 second) and modern broadband connections will often have an average latency of anything from around c.4ms to 40ms (what is normal for your connection will depend on lots of different factors).
A faster score (shortest time) is best for latency, although this can be affected by various things, such as the performance of remote internet servers, the connection technology being used, network congestion at your ISP, peering / routing problems and the setup of your own home network etc. But consumers with a modern Fibre-to-the-Premises (FTTP) based broadband connections can usually already expect some of the best latency times (often c.2-10ms).
Advertisement
However, Vodafone and other operators have long been working to introduce L4S technology, which could enable their networks to simultaneously maintain high throughput and low latency. L4S is based on the knowledge that the root cause of queuing delay (i.e. when packets wait idly in buffers across the network, for instance in routers and modems, before being forwarded) is in the capacity-seeking congestion controllers of senders, not in the queue itself. L4S thus aims to transition away from congestion control algorithms that cause substantial queuing delay and instead adopt a new class of congestion controls, which seek capacity with very little queuing.
The good news today is that they’ve progressed from last year’s lab trials and have now trialled L4S on a live commercial FTTP fibre broadband network (end-to-end) for the first time. The trial took place in Türkiye (Turkey) and reduced latency on multiple customer lines whilst using video conferencing and cloud gaming applications.
The increase in round-trip delay, with the customers line heavily congested, was said to have been reduced from 220ms to just 4.7ms.
Alberto Ripepi, Chief Network Officer of Vodafone, said:
“We are raising the bar on quality for fibre broadband by significantly lowering latency. Our world-first trial underlines our commitment to innovate and deliver an enhanced service to customers across Europe.”
Bjorn Capens, Vice-President Fixed Networks Europe of Nokia, said:
“Nokia continuously strives to increase the performance of its fibre solutions through innovation, and this trial puts the L4S internet protocol – pioneered by Bell Labs – into practice in a real-world fibre network, including the in-home Wi-Fi.
Innovating for and with a leading operator like Vodafone is paramount for a company like Nokia, as well as providing superior latency with minimal latency variation on an end-to-end fibre and in-home Wi-Fi network.”
ISPreview queried when we might see this being deploying on Vodafone’s network in the UK, although the operator is currently planning to conduct more tests before looking to deploy the technology across its European markets (focusing on driving common standards and vendor support). But they do still expect it to be “adopted over the next few years“.
Advertisement
We should point out that L4S is a technology neutral standard, so while the trial was conducted on FTTP, it could also just as easily have been deployed on 4G or 5G mobile etc.
Advertisement
My latency when pinging a popular DNS service (Cloudflare 1.1.1.1) has dropped from ~9ms to ~6ms following the switch from FTTC to FTTP at the same location. I’m guessing this is fairly typical, but it’s obviously a good thing. It’ll be very interesting to see a technology able to bring this even lower, but given the number of links in the chain (in my case, 4 active devices before it even hits the physical fibre), I’m curious to know more about how and where this tech is deployed.
The focus of L4S is on bringing latency down when your broadband connection is heavily congested, rather than improving unloaded latency.
The work going on around optical computing will deliver significant step improvements in latency figures.
It is still early days but significant progress has been made recently.
Welcome news.
Why not roll out fq_codel?
If it’s good enough for the mainline Linux kernel to use as the default for network interfaces, surely it’s good enough across the board.
L4S has quite a few issues and backwards compatibility problems:
https://github.com/heistp/l4s-tests/blob/main/README.md
If only their equipment could support this 😉 What they’ve done was basically adding a prio/tos on icmp and maybe most common ports used by video games. When I signed with OFNL, my latency to Linx/Lonap was 4.5-5ms, now it is 6-7ms.