Okay, I’m actually onboard with those things. It’s usually a cable with a single wall plug on one end and four C13 plugs at the other. So you could plug in two monitors, a desktop PC and a printer, say, with just one socket. They’re a lot neater than having a whole power strip and four cables. I’ve also seen ones that split one C14 into four C13s but I’m not as sure how I feel about those.

Given the 2.5Gb port also supports PoE in, I think the idea is that you can plug this into a 2.5Gb PoE port on a seperate managed switch and that’s the only connection you need; that’s certainly how I would use it. WAN connections could be plugged into that switch, along with the APs, user devices, servers, etc, with them seperated using VLANs. Assuming everything was gigabit except for that 2.5Gb link to the OpenWRT Thing™, you’d be hard-pressed to saturate that 2.5Gb port and you’d still have the gigabit port completely free for… whatever.

I think you might have missed the point: with a managed switch that 2.5Gb port can be used to handle multiple WAN and LAN connections simultaenously. My home network includes two WANs and six LANs split purely by VLAN tagging and that 2.5Gb connection should handle all of them just fine.

It’s full duplex so it’s 2.5Gb each way simultaneosly. Most NICs support half-duplex but I don’t know of any good reason to use that. I used to have a BananaPi based router that could comfortably saturate it’s gigiabit interface. I assume there’s some kind of offloading going on.

It happens.

Source: https://www.youtube.com/watch?v=hqIYmYBNYxE&t=1646s

Most of those run OpenWrt or PfSense. Assuming the hardware is well-supported by the open source software it runs, there’s a argument to be made that there’s no difference. There’s always the risk of them using some weird chipset that won’t be supported in a year’s time. The only difference is that the OpenWrt One is specifically designed for OpenWrt with well-supported hardware.

I also wanted to chime in with the perennial point that while this device is a pure expression of the OpenWrt project, they also support hundreds of other devices including, amazingly, a number of large switches, so if you wanted to ditch the separate route appliance altogether you could get all the features with only switch hardware.

Exactly this. With VLAN tagging you can plug that single 2.5Gb connection into a 48-port managed switch and effectively have up to 47 different NICs if that’s what floats your boat. They’d all share the 2.5Gb but that’s still more than a lot of small networks need.

The LAN and WAN ports aren’t labelled as such on the device and can be configured to do anything. The 2.5Gb port can also be used to take in PoE so for a lot of people - myself included - this will be the only port that’s actually used, or at least the port that will be used the heaviest. The reason, I think, that it’s configured as WAN by default is so that the LAN port can be used to plug a laptop in directly without disconnecting the whole network.

Devices should be flexible.

Agreed. It’s a pity, then, that no-one has invented a single port that can replace USB-A, DisplayPort, HDMI, propriatary power connectors, Thunderb… oh… wait…

The USB4 protocol can handle 160Gb/s split asymettrically (so, say, 120Gb/s out, 40Gb/s in), wheras the upper limit for DisplayPort’s highest bandwidth mode, Quad UHBR 20, is 80Gb/s in one direction. So you can saturate your DisplayPort 2.0 quad-channel with more than enough bandwidth to power three 10K 60Hz 30-bit (i.e. very high-end) monitors in DSC mode, and still only be using half the bandwidth of USB4, all using a single cable which I can also use to charge my earphones.

If I had an EV it would make a huge difference. One of the really, really cool features of the Octopus integration is that you can create a binary sensor that triggers for the cheapest hours (consecutive or non-consecutive) between a set time period, so you could create an automation that works out how many hours your car needs to charge at, say, midnight, and have it only charge for the cheapest hours between then and 8am.

In my old place I actually did this: replacing my fridge and freezer’s thermostats with an ESP Home controlled relay and thermometer. This place has a fancy integrated unit that I don’t want to play with too much.

In fairness, most people here pay about 24p/kWh which is about 32¢/kWh and on my tariff sometimes prices will peak at 45p (about 55¢) so we’re not too far behind!

I’m surprised, to be honest, since I worked out a while ago that if I had an EV - even something like a Zoë doing 20 miles a day - I’d be saving a lot of money, but if you’ve done the maths then fair enough. Keep in mind, though, that since you’re alreay with Octupus you could just switch to Agile for a month and switch back if it doesn’t work out.

Dingdingding! Correct. For the chepest two hours a day (or any time cost is negative) Home Assistant gives Portainer a kick and I sail the high seas. Whenever costs are negative I saturate my servers with BOINC CPU-heavy workloads like ClimatePrediction, Rosetta@Home, LHC@Home and World Community Grid.

Whole-home metering is done with a Home Mini, a device that Octupus gives away to any customer that asks nicely and provides real-time data. For devices with plugs is mostly LocalBytes smartplugs or similar. For the heaters, well, hypothetically, that would require installing something like a Shelly PM Mini Gen 3 inside the wall box behind the heater without asking the landlord’s permission.

I’m currently paying £50/mo and that’s with credit building up on my account. My initial investment in HA has paid for itself many time over.

Yup! No EV here, sadly, and I live in a flat but I’ve got storage heaters and a big hot water tank. I’ve got an incredibly janky template sensor that works out how many hours of heating I need for each room based on the weather forecast and an automation that activates the heaters for that many hours a day at the cheapest times. It can also turn the heating on when the price drops below a certain threshold, currently 0p.

It’s Home Assistant. If you’re not familiar, it’s an insanely powerful open-source platform for self-hosting your smart home. It can handle heating, lighting, security, charging, presence, all sorts. I highly recommend it if you have any smart home kit at all. But it is a bit of a rabbit hole.

The data you see there is being automatically imported from Octopus and Home Assistant uses it to calculate the best time to run the heaters; it even takes the weather forecast into account to make sure that the walls are warmed through so the heaters don’t need to run when energy is expensive.

The 22:30 is highlighted because it’s the cheapest rate for the period.