Original article ISPreview UK:Read More
A small but interesting development today. Apple released the latest iOS 27 Beta 6 software to developers yesterday and those who have updated their iPhone’s to the new firmware are finding that it finally shows a “5G+” symbol (or icon, if you prefer), when in areas covered by a 5G Standalone (5GSA) based mobile broadband network. (more…)
Original article Total Telecom:Read More
Provided by: Belden
For over a decade, Power Usage Effectiveness (PUE) has been the go-to metric for evaluating data center efficiency. As artificial intelligence (AI) workloads scale rapidly, however, relying so heavily on PUE has run into a problem.
Unfortunately, PUE’s facility-wide perspective provides little insight into how efficiently an AI workload is being executed. As AI workloads continue to demand growing amounts of power, the data center industry is pivoting to a more granular metric – energy per inference.
Optimizing for this metric requires scrutinizing every layer of the hardware stack. While GPUs and advanced cooling systems receive significant attention, the optical interconnects that move data within clusters are quickly becoming a strategic concern.
In legacy data centers, the power draw of a single transceiver has always been a topic of importance, but less so than today. Modern AI fabrics have changed the conversation.
A high-performance GPU server can rely on ten or more optical transceivers for communication across spine, leaf, and server switch architectures. As data rates shift upward to 800G and 1.6T, the energy cost of moving bits begins to rival the cost of processing them. At this scale, optical design choices begin to influence energy per inference in a measurable way.
Linear Pluggable Optics (LPO) simplify the transceiver by removing the most power-intensive parts – the Digital Signal Processor (DSP) and Clock Data Recovery (CDR). In standard full re-timed optics (FRO), the DSP alone can account for roughly 40% of total optic power consumption.
Instead, LPO shifts signal conditioning responsibilities to the host switch silicon, specifically the SerDes (Serializer/Deserializer). This handoff results in a leaner optical module that slashes the power draw per 800G link from around 13W-16W to 7W-9W. Lower transceiver power means less heat generated at the switch faceplate. Less heat means reduced cooling demand across the room and less pressure on HVAC systems.
Additionally, by bypassing the DSP processing cycle, LPO reduces latency from around 100ns to less than 10ns. That not only improves the responsiveness of distributed AI training and inference but also permits infrastructure to deliver more useful compute work per watt-hour.
Crucially, LPO modules interoperate with standard FRO, allowing operators to maintain hybrid environments and pursue a phased transition within the same network fabric.
Despite the immense benefits of LPO technology, its successful deployment depends on a few, key hardware conditions.
As AI clusters move toward 1.6T interconnect speeds, pressure to improve efficiency will continue to rise. A growing number of industry analysts see LPO as a practical near-term deployment solution that will enjoy significant, consistent growth through 2033.
In the AI era, improving energy per inference means looking beyond the GPU and examining every watt consumed across the network path. Linear Pluggable Optics are not a universal replacement for traditional re-timed modules. But in short-reach, high-density AI environments, they can play an important role in reducing power draw, lowering thermal load, and improving compute efficiency. For those building at scale, the optical layer is no longer a background consideration. It is now the next fundamental frontier in the quest for more sustainable AI.
Subscribe to the Broadband Communities newsletter!
The post Why Your Optical Layer Is Now an Energy Per Inference Strategy appeared first on Total Telecom.
Original article Total Telecom:Read More
At 03:00 next Sunday, someone will be on shift, responsible for your network. It is worth asking who that actually is, and whether they can read a BGP table or only acknowledge an alarm. For a growing number of UK fibre operators, the honest answer is the second one, and for 128 of the 168 hours in a week there may be no answer at all.
That question used to be survivable, because the market rewarded building, but that era is closing. Ofcom puts full fibre at more than three-quarters of UK premises, and AlixPartners estimates nearly half of UK fibre operators must refinance by the end of 2026, at rates far above what they borrowed at. Lenders no longer fund premises passed. They fund take-up, revenue and cost discipline. The sorting criterion has moved from how much network you built to how well you run it.
The gap every operations lead already knows about
Most operations leaders at altnets know exactly where their gaps are, overnight coverage is thin, the inventory has drifted, the change process depends on one person being awake. What they lack is the headcount economics to fix it.
Covering one 24/7 seat sounds like simple arithmetic, 168 hours against 40-hour weeks suggests five engineers. Anyone who has built a rota knows that people need rest after nights, and someone is always on leave, off sick or working on their notice. A sustainable team takes six engineers, or five plus a lead, and it cannot be six juniors because a seat that can resolve incidents at 03:00, not just acknowledge them, needs a mix of first-line and senior engineers, which in the UK means salaries from around £33,000 to £50,000 and beyond, per public salary data. That puts the team behind that one seat at roughly £250,000 a year in base pay, and £350,000 or more with employer costs, shift premiums, recruitment, training and tooling. All of it buys one person watching at any given moment. In a consolidating market, closing this gap by hiring is the most expensive answer, competing for the same scarce engineers everyone else is chasing.
What we find inside real networks
ITcare runs managed NOC and network engineering for ISPs, data centre operators and hosting providers across North America and Europe. When we take over or extend an operator’s network operations, the pattern is consistent, the inventory is wrong in ways nobody has quantified, monitoring has either no visibility or produces a ton of alarms nobody has triaged in months, and the topology lives in one senior engineer’s head. None of this reflects badly on the teams involved. It reflects an operations function built during the build era, now carrying an operate era load it was never staffed for.
The specialist model exists for exactly this, a dedicated team of service provider engineers takes the 24/7 load, the change discipline and the escalations, working as a standalone team or an extension of the operator’s own team, and usually for less than what staffing a single 24/7 seat in-house would cost. The in-house engineers stop firefighting and go back to the work that grows the business.
The AI claim, stated honestly
Every stand at Connected Britain will say AI this year, so let me say it precisely. We build and operate HORA, an AI network operations platform, and we ran it inside our own NOC, breaking things in our own lab, before any customer touched it. That made us conservative in a specific way.
AI earns its keep on the read side. HORA brings metrics, logs, flows, BGP state, discovered topology and the source of truth into one place, and an AI engineer correlates across all of it to get a human to root cause in minutes. That is why a lean team can operate more network, and why, after a merger leaves two networks and two half-accurate inventories, automated discovery turns an integration plan into an integration. The write side is different. We keep AI off it by architecture, it proposes, a human approves, and what touches a device is deterministic automation a person wrote and can audit. Any AI pitch that skips this distinction is selling you the demo.
HORA correlating live network state to root cause.
The operate era
The strategic question for every operator heading into the autumn is not how much network you have, but whether the way it is operated would survive due diligence. If the honest answer is no, that is fixable, and it costs less than you think.
We will be at Connected Britain on 9 and 10 September, stand 140. Bring us your hardest operational problem, the inventory nobody trusts, the alarm queue nobody reads, and we will tell you on the spot how we would approach it. To set up a meeting ahead of the event, write to an****************@****re.net.
Andrian Visnevschi is CEO and CTO of ITcare and holds JNCIE-SP #2975. ITcare runs managed NOC and network engineering for ISPs, data centre operators and hosting providers worldwide, and builds HORA, an AI network operations platform.
The post The build is over, but can you operate what you built? appeared first on Total Telecom.
Original article ISPreview UK:Read More
Hampshire-based alternative network toob, which has deployed their own gigabit speed full fibre (FTTP) broadband network across South England and also harnesses CityFibre’s network in other UK areas, has suffered a blow today after investor INPP “elected not to commit further capital” to the provider and handed most of its original £24.1m investment off to debt holders. (more…)
Original article ISPreview UK:Read More
Managed ISP ASK4, which is headquartered in the UK and delivers gigabit-capable broadband to students and residents (renters) in multi-tenant buildings across the UK (as well as parts of Europe), has announced the completion of a £170 million refinancing project with new capital expenditure facility to help “accelerate growth“. (more…)
Original article ISPreview UK:Read More
The UK government’s proposed plan to switch-off Digital Terrestrial TV (DTT) signals in either 2034 or 2044 as part of a wider change to move toward broadband-based TV streaming (here), such as via the new Freely service, arguably suffered a bit of a reality check last week after that very same service experienced a major outage. (more…)
Original article ISPreview UK:Read More
A business update from technology provider Dotlines UK has today revealed that they’ve signed a 5-year contract with retail broadband provider Olilo, which will see the ISP deploy Dotlines Catena Cloud platform, which is an all-in-one system that’s designed to help internet providers to “seamlessly sell, deliver and manage end-to-end telecoms services“. (more…)
Original article ISPreview UK:Read More
Network operator EE Mobile is preparing to launch a new “Full Works Plus” airtime plan in the near future (probably the next month or so), which will cost £5 extra per month than their regular “Full Works” plan and introduce enhanced 5G+ (5G Standalone) mobile broadband network slicing via a new ‘Fast Lane‘ feature. But there’s a big caveat. (more…)
Original article ISPreview UK:Read More
The Government’s Department for Education (DfE) has launched an early market engagement exercise to identify organisations that might be interested in supplying the second phase of their gigabit broadband roll-out programme for Schools, which has been allocated £20m of public funding to reach around a further 800 schools across England. (more…)
Original article ISPreview UK:Read More
New student customers of mobile operator giffgaff, which is controlled by Telefónica UK (part of the VMO2 Group) and uses O2’s virtual network platform, may like to know that the provider recently introduced a new SIM that gives you 100GB (GigaBytes) of mobile broadband data and other features for just £10 per month. (more…)