I think I mentioned in a previous post that some manufacturers (such as Intel and Cisco) can demand the SFP/SFP+ module is branded by them or they simply won’t work (they use a manufacturer code which the SFP needs to supply).
For example, Finisar make SFP modules which, while recognised by Intel ethernet boards - and, otherwise, identical - will not allow themselves to be be enabled unless they identify themselves as Intel modules.
I also have experience of SFP’s not fully working with other manufacturers’ gear (such as Cisco SFP’s with Ubiquiti equipment not negotiating the speed at 10Gbps until a hard reset, meaning the connection won’t work on a reboot).
This isn’t generally a problem on switches made for the audiophile market as they don’t tend to use any locks (meaning any SFP should, in theory, work), however there can still be compatibility issues I just mentioned. TBH Most of this works as intended, though.
I presume we are talking about Finisar here not Finisair.
I need a good website anywhere in the world which will help me navigate their model numbers as all the ones I’ve come across so far are 10G or RJ45 or Multi Mode… I won’t necessarily buy from there but need a decent map of the territory!
I have used the website I pointed out earlier to narrow down the extensive Finisar range (it looks like they list 198 models!), with the following parameters selected:
Manufacturer = Finisar obviously
Connector = LC ( not SC, RJ-45, MPO, SFP28)
Form type = SFP (not SFP+ or several others)
Media = SMF (not MMF)
Speed = 1000Base (not 100GBase, 10GBase etc)
Transceiver category (unticked=all)
I also left all of these unticked=all too:
Type, Wavelength, Country, DOM, Environment, ITU Spacing, ITU Frequency, ITU Wavelength
The one parameter which made me pause was Transceiver type. There are a dozen or more listed, so I did some research and found out what some (CWDM, DWDM, ZX etc) but not all mean. I eventually ticked only LX. Am I right to do this?
I bought two* Finisar FTLX1472M3BTL (1310nm). Various options were recommended online, however trying the top three or four recommendations, my own ears preferred these.
*I actually ended up paying for four as they processed my order twice, but wanted me to pay the return shipping and insurance costs from the UK to the US to get my money back, which rather defeated the object. I’ll need to find a way to incorporate two more into my system!
I’m pretty certain mine are running at 1Gbps anyway (the switches only support 1Gbps), but I couldn’t tell you whether going for one of their 1 or 10Gbps SFP’s makes a difference.
I doubt it does, although I believe more is done to prevent jitter on the ethernet circuit on the 10Gbps modules. Whether that makes a difference is debatable.
As Jeremy mentions, some Switch/Router vendors will reject non-appropriately coded SFPs, though most consumer-grade Switches will just ignore the coding.
The coding is carried on EEPROMs within the SFPs, they’re multi-byte fields for things like the vendor name, ID, part #, serial #, SFP version numbers, etc. Depending on the SFP, the coding can actually be reprogrammed “in the field” with specialised EEPROM reprogrammers.
Because large Customers have enough sway to strong-arm Vendors into letting them use (significantly cheaper) 3rd party SFPs, most vendors, including Cisco, have CLI commands that enable the Switch/Router to ignore mis-matching vendor-coded SFP, and just work. The commands do vary depending on the platform/version of Cisco IOS, but the following usually works on Cisco Catalyst Switches;
(config)# no errdisable detect cause gbic-invalid (config)# service unsupported-transceiver
FWIW These are the two other modules I found recommended online:
StarTech SFP1000ZXST (I tested with and without attenuators)
Cisco SFP-10G-LR
I tested these alongside the Finisar above and, to my ears, the Finisar sounded better. Others have said it is not possible for the SFP modules to sound different. While I don’t wish to continue that debate, I thought I’d include this information for those who want to experiment themselves.
I will caveat this, however, with the fact that I usually go back and run the listening test a second time many weeks later to see if I still hear a difference. I haven’t had the time to do this so far…
