https://www.ui.com/us/en/integrations/accessory-tech/sfp-wiz...
Previously seen: https://news.ycombinator.com/item?id=45732874
Especially handy for specific Intel NICs where they refuse to link up if the module isn't in the driver-allowed list and those modules are hard to come by
You are correct that 10GBASE-T really shouldn't be the default choice, fiber and DAC both have advantages over it. But compatibility is important, and there are a lot of situations where 10GBASE-T is just more convenient.
Regular people also are not buying DACs.
If you are in the line of work where you need to know what SFP is and the difference between DAC and Optical, a quick "what's OM3 vs OM5 and when do I use either?" to your favorite LLM/Search engine will get you sorted.
Also personally, if you can get away with a copper DAC, I would rather use that instead of fiber because you don't need any special modules.
10baseT (!0Mbps) came out in 1990 (there were non-twisted pair earlier versions). "Fast Ethernet" (100Mbps) came out in 1995. Copper 1GbE came out in 1999. Copper 10GbE came out in 2006. Ethernet seemed addicted to 10x'ing every version and 10GbE is really where everything fell apart. Or at least, it's where it got hard. We never really got mass market 10GbE. The controllers were too expensive. The cable requirements were quite high. And heat was an issue.
1GbE really was fast enough and 10GbE was a massive jump that I even remember thinking at the time that there should've been intermediate steps, which is what happened in 2016 with 2.5GbE and 5GbE.
Now compare to Thunderbolt, introduced in 2011, which has completely surpassed Ethernet bandwidth, in part by putting chips in the cables, but of course the big difference is cable length. A copper cat 6/7 cable can get to ~100 meters, which is also why the power is so high: attenuation.
but I guess my point is that 10GbE over copper was a mistake. We'd reached the point where you really had to swap over to fiber.
jauntywundrkind•1h ago