I'm not sure if the 93180YC-FX is fully 802.3by-compliant, but it does support RS-FEC.
25G Ethernet requires forward error correction (FEC) everywhere but 1 & 2-meter passive copper cables. There are two 25G Ethernet FEC algorithms:
- FC-FEC (aka "FireCode" or "BASE-R" or "Clause 74") offers simple, low-latency (< 100ns) protection against bursty errors, but is not so good for random errors. It's used on 3 and 5-meter passive copper cables, and on active optical 25G cables up to 10 meters. (it's also used on every 100G interface)
- RS-FEC (aka "Reed Solomon" or "Clause 91" or "Clause 108") offers better error protection but adds latency (~250ns). It is required for the 25G MMF transceivers (Cisco p/n SFP-25G-SR-S) for distances up to 100 meters. I suspect it might also be needed on active optical cables over 10 meters long (if Cisco sold them).
Every 25G device supports FC-FEC, but RS-FEC is newer and more rare. E.g., the Nexus 9300-FX is Cisco's first switch to support RS-FEC. QLogic's QL4521x doesn't do RS-FEC, but their QL41212 apparently does.
Product incompatibility will be a minor headache, but autonegotiation could be what we curse in the future! The IEEE struggled with adapting autonegotiation for FEC, and many early 25G products may not understand the negotiation from the final 802.3by draft. I expect a lot of vendor interop issues with FEC negotiation, and a lot of frustrated end-users who resort to hardcoding it.
[Update Nov 2018] Do check out Cisco's new SFP-10/25G-CSR-S transceiver. With two, it claims 70/100 meters (OM3/OM4) with FC-FEC and 30/50 meters without any FEC. It also interoperates over regular distances with standard 10G and 25G-SR transceivers.
