Lots of replies, but not a lot of product recommendations.
Netgear GS105 5-port Gigabit Ethernet Switch - $40
TP-Link TL-SG105S 5-port Gigabit Ethernet Switch - $20
There’s no particular reason to choose one over the other. The TP-Link is cheaper, but some people like the darker color of the Netgear. Not sure it’s worth twice the cost though.
None of the major ISPs offer SLAs (service level agreements) with throughput guarantees for residential service. You’ll hear anecdotes from people here about their performance, but this is specific to their market and their specific network segment. You can’t make the leap from those anecdotes to “service type X is better” because it’s not the service type that’s delivering more throughput; it’s the ISP network design.
Here’s the thing. All residential internet works over shared medium networks. The wire or fiber that goes back to the distribution node is shared amongst your neighbors. For fiber, it’s called a PON (passive optical network). For cable it’s called DOCSIS (please don’t at me if you’re an engineer, I’m generalizing). These networks use a shared physical layer to supply data to the last mile.
So your actual throughput is usually limited by two factors:
ISPs routinely oversell the throughput available on these networks. It’s how they turn a profit. Getting good speeds is a matter of finding the provider with the least over-provisioned network in your area. That last part is important. An ISP can be good in one region and lackluster in another.
All that said, fiber networks tend to be newer, so it takes time for ISPs to sell them out. For example, if the cable provider in your area had a near monopoly, then the local phone provider rolled out fiber, it’s going to take time for people to switch. You can take advantage of that by adopting early. Just understand that your network performance will degrade as more people switch and you end up with more network contention.
The answer comes down to cost. LiFePO4 batteries have enough capacity that they can be held at 80% and still beat SLA on capacity. The problem is that LFP wins on gravimetric and volumetric energy density; two metrics that aren’t particularly important in the context of a UPS. It’s a large brick that sits on the floor.