I disagree - SerDes have nothing to do with the actual link establishment between 2 modules, it kicks in only after. SerDes 8B/10B is the reason why 1,25Gbps SFP only handles 1Gbps of actual traffic
1.25 GBps is just the electrical bandwidth of the tranceiver electronics.
1) RJ45 modules can be 10/100/1000 or 1000 only if trying to connect to 100Mbps device one will work other will not. Board will remain the same
A RJ45 SFP may be a special case, but on Cisco equipment it only handles 10/100 links speeds if the host port you plug it in supports this. Also the SFF MSA doesn't cover RJ45 tranceivers.
2) you can plug 10Gbps only SFP+ transceivers on RB2011 and they will establish link and work. board will be able to supply only 1Gbps traffic, but link itself will be 10
You can plug them in and they will establish a link, but I doubt this link runs on 10GBps. Its just the SFP which supports a high electrical bandwidth, which isn't used in this case.
Try plugging in the other end into a 10GBps only port on a Cisco Nexus for example.
similar to previous -there are optical modules that support whole range 155M~2,63Gbps - my guess they will use some kind of auto-negotiation to establish the link and link speed.
Sure they support electrical bandwidths up to 2,63GBps why shouldn't they supprt lower bandwidths.
The SFP MSA defines just two differential transmit and receive connections (TD+/- and RD+/-), some status lines(Transmitter fault i.e.) and an onboard EEPROM chip to identify the module and its capabilities like max. speed, distance etc. It doesn't define any bit rates or protocol or encoding(just supported encodings). While there are SFPs which are more then just dumb tranceivers, like SFPs which include full POS support and emulate
an ethernet link on the eletrical interface, this isn't true for just standard SFP optical tranceivers.
You may have a look at SFF INF-8047 here ftp://ftp.seagate.com/sff/INF-8074.PDF