Yeah, i get your point.
It’s not unusual to have manufacturing partners who just happen to produce “a little extra” on the side. As long as there is no Intel logo, it’s even potentially legal though I’d be surprised if Intel officially sanctioned it.
Similar practices abound in the automotive world. OEMs like Bosch will make custom parts for VW, BMW, etc that are available in a branded (ie with car mfg logo) and OE supplier only version - the latter being much less expensive. Sometimes it’s as simple as the car brand logo being scraped off the part at the factory.
See m539 restorations YouTube channel or take a gander through the selection of parts at pelican parts. A branded Bosch windshield wiper motor is available there for my car at $350 or $105 w/o branding. It was also available used on fleabay for $28 from a wrecker, so that’s where I bought it.
Bottom line, test the hell out of it and if it passes and performs consistently at 10GbE, who cares?
1 Like
Wow
Everything matches as per the image presented on the Intel’s official website and the datasheet. The serial number on the optics label matches with that on the box and also via the ethtool and the ethtool shows the Vendor name is Intel Corp. The only thing missing is the Validation label at the back of the transceivers.
As mentioned in the previous post, one my friend who is a homelab lover, mentioned that not all the batches will have the validation label as its not Intel who manufactures themselves but other semiconductor corp. does for Intel and other brands. Although i’m not sure about this particular information, but I think he is saying the same thing what you said i.e. producing a little extra on the side 
Another friend of mine told me that he pulled off some of the modules and the NICs from the running servers and some of the labels were torn off during the pulls and that the Yottamark/Brady ID stickers are on the Capacitors side, not the plain surface and they don’t hold/stick very well.
Hmm. Makes sense friend.
Yeah, i agree with your point but the motor without the branding is really a genuine part? That’s the main question. I mean, yeah, both the motor will work. It could be also a possibility that the genuine (one with label) and sourced officially, can fail earlier than the counterfeit one. But it could affect the performance or may fail way earlier than the genuine part.
For example, talk about HBA Cards. The counterfeit one also uses the same LSI chip, but the build quality is way much different, like selection of the other parts such as capacitors, PCB quality, heatsink, etc.
The real question is what about the OEM parts? As one can manufactures the counterfeit parts (NIC and modules) with the genuine part number. So, how does that verify if its a genuine part, unless the customer bought it from official sources like Dell, HP, Lenovo, Inspur, IBM, etc.
Yeah, really not much of worry 
Thank you for sharing your opinion!
So, do they work?
1 Like
Yes, it works but i’ve not been able to fully test like loss of packets or checking constant speed during the transfers. I had to work on something so i just quickly checked if it shows the link and matched serial on all the three things (box, module and the ethtool)
Yes, BOSCH is the OE supplier in both cases. As a consumer you are given a choice between a BOSCH made part w/o the car OEM logo and a BOSCH made part made with it.
Ultimately, the only difference is the logo but if you’re a licensed car repair shop, you’re supposed to use OEM parts w/the logo. Ditto if you’re preparing your car for the Concourse d’elegance in Pebble Beach.
But m539 restorations makes a point of showing off the OE suppliers re replacement parts while saving his viewers money w/o compromising on quality. For injection-molded parts that means the OE supplier has to scrape off / grind off the OEM logo that otherwise would be part of the injection molding.
For other parts like the pulley wheels, they are available in all diameters, etc. directly from the OE supplier. Same quality, etc. Only a few parts are visibly different if you know what to look for. For example, HELLA makes some headlight assemblies where the interior reflective bits are slightly different because the branding is so prominent. But the functionality is the same.
In the case of the transceiver, it’s pretty difficult for Intel to control what the OE supplier does and doesn’t produce. They don’t make the transceivers in house, they farm it out to the Foxconn’s of the world.
1 Like
This was good information. Thanks for sharing.
As far as i’m aware, the Extended Temperature modules are manufactured by Hisense Broadband for Intel and its Finisar mostly along with Broadcom who manufactures commerical one. The only exception is the 100GbE module (Commerical), which is manufactured by Silicon Photonics (I guess its the Intel) and i have seen a lot of broken ones.
I’m waiting on the rest of the parts to arrive and then i can test it easily. I just hope i did not get counterfeit modules. They costed me over 1K+
1 Like
Then you must be buying a lot of modules? Used intel from reputable sellers on eBay are less than $40 last time I checked for 10GbE. The 25GbE modules you’re after are maybe twice as expensive and available used from seemingly reputable resellers also.
I wouldn’t be buying new Intel for my home use. It’s too expensive given that 10Gtek makes great new, but Intel-compatible transceivers at the eBay price of genuine used Intel. I prefer genuine used but use both types of transceivers in my equipment.
They were 16 units.
Yes, that’s right!
Neither me. xD
Yes, i always prefer genuine. Never used the compatible one. But this is also my first time and i’ve begin my journey with optics
1 Like
There are option, I will also 2nd 10GTek, so far “knock on wood” 2 receivers and several DAC’s I have for them have been good.
1 Like
I’ve had my fair share of issues related to transceivers but most of them had to do with configuration unhappiness, not hardware. Sometimes, it is far harder to get something hooked up as desired than one thought it should be.
@stux has a great illustration of the logical two-step dance one has to perform to enable bridge mode on the TrueNAS SCALE platform for jails. He thought it out well, allowing mistakes along the way to get rectified vs. trying to do it all at once, failing, and ending up have to nuke all transceivers from the console…. again.
I found optical in a SOHO setting to be surprisingly easy once config issues are addressed. I wish OEMs were clearer re: vendor lock (since this is a bogeyman of sorts that sows doubts when something isn’t working as intended).
The funny short-sighted effect of vendor lock is that it drives counterfeits to a higher level.
5 Likes
Maybe the vendor lock status should be put into the TrueNAS documentation or the Networking Guides?
Q/SFP/+/28 modules are, and have always been, a s*** show. But trust me when I say, it’s so much better now than it has ever been.
Living in the datacenter for a long time, I’ve developed several bad habits, referring to them all as gbics rather than tranceivers because my brain has been poisoned by papa Cisco. I was actually responsible for a few 6509Es and Nexus 7000s that used these god-forsaken things called “XENPAKs” as recently as 2020.
Fast forward to today, where everyone is more or less using the same standardized “small-form factor” variants, the waters became more murky than you’d think despite the standardization in form factor. When things were physically different, you at least knew what you got was the right thing.
I’ve used many 10G-BaseT adapters that outright lie that they are an SR optic and even spit out hot garbage lies about optical TX/RX and other values (they aren’t optics!). Despite this, they actually work just fine in most cases.
fuscoastd0ca# show interface transceiver details
Ethernet1/1
transceiver is present
type is 10Gbase-SR
name is FIBER MALL
part number is SFP-10G-T-CI-80m
revision is A
serial number is MT221410154
nominal bitrate is 10300 MBit/sec
Link length supported for 50/125um OM2 fiber is 82 m
Link length supported for 62.5/125um fiber is 26 m
Link length supported for 50/125um OM3 fiber is 300 m
cisco id is 3
cisco extended id number is 4
SFP Detail Diagnostics Information (internal calibration)
----------------------------------------------------------------------------
Current Alarms Warnings
Measurement High Low High Low
----------------------------------------------------------------------------
Temperature 29.69 C 80.00 C -10.00 C 75.00 C -5.00 C
Voltage 3.30 V 3.59 V 2.90 V 3.50 V 3.00 V
Current 6.00 mA 15.00 mA 1.00 mA 13.00 mA 2.00 mA
Tx Power -3.00 dBm 0.49 dBm -7.30 dBm 0.00 dBm -6.30 dBm
Rx Power N/A 0.49 dBm -11.93 dBm 0.00 dBm -10.91 dBm
Transmit Fault Count = 0
----------------------------------------------------------------------------
Note: ++ high-alarm; + high-warning; -- low-alarm; - low-warning
So quite literally, there’s no way to report any of this information in an accurate way in TrueNAS, because modules are known to lie.
For what it’s worth, I have another 10G-base T module in this same switch that literally won’t work because IT DOESN’T lie, and my Nexus vendor locks it. I’ve had similar problems with HP Procurves in the past, so it’s not even JUST Cisco.
2024 Jul 14 16:01:39 fuscoastd0ca %ETHPORT-3-IF_XCVR_ERROR: Interface Ethernet1/47, platform doesn't support SFP-10G-T-X
2024 Jul 14 16:01:39 fuscoastd0ca %ETHPORT-5-IF_DOWN_ERROR_DISABLED: Interface Ethernet1/47 is down (Error disabled. Reason:Unsupported media-type or configurat
ion)
Manufacturers of “compatible” SFPs have been forced into this awful lying behavior by vendor locks, as said above. However, from the technical end, there’s still real concerns about compatibility, stemming from power-draw and quality of components.
I’ve had a Mikrotik crs305-1g-4s+in switch that would do all sorts of whacky behaviors and shut off if I ACTUALLY used all of the ports because the transceivers I used were drawing too much power. See Building a Lab Part 2 Building Our Network - Page 3 of 4 - ServeTheHome
So it’s not arbitrary that vendors like Cisco do this. It’s just complicated, and it’s not isolated to 10G BaseT. DAC cables are also notoriously problematic at times. Same switch, with an OEM Real Cisco 40 gig DAC cable going to my pfsense router in an Intel XL710. The port was randomly flapping every few days, until suddenly it just stopped working. My Intel XL710 card no longer works at all, the DAC cable killed it.
2024 Jun 25 15:47:59 fuscoastd0ca %ETHPORT-5-IF_DOWN_LINK_FAILURE: Interface Ethernet2/12 is down (Link failure)
2024 Jun 25 15:48:02 fuscoastd0ca %ETHPORT-5-SPEED: Interface Ethernet2/12, operational speed changed to 40 Gbps
2024 Jun 25 15:48:02 fuscoastd0ca %ETHPORT-5-IF_DUPLEX: Interface Ethernet2/12, operational duplex mode changed to Full
2024 Jun 25 15:48:02 fuscoastd0ca %ETHPORT-5-IF_RX_FLOW_CONTROL: Interface Ethernet2/12, operational Receive Flow Control state changed to off
2024 Jun 25 15:48:02 fuscoastd0ca %ETHPORT-5-IF_TX_FLOW_CONTROL: Interface Ethernet2/12, operational Transmit Flow Control state changed to off
2024 Jun 25 15:48:02 fuscoastd0ca %ETHPORT-5-IF_UP: Interface Ethernet2/12 is up in mode access
2024 Jun 25 15:48:35 fuscoastd0ca %ETHPORT-5-IF_DOWN_LINK_FAILURE: Interface Ethernet2/12 is down (Link failure)
2024 Jun 25 15:48:38 fuscoastd0ca %ETHPORT-5-SPEED: Interface Ethernet2/12, operational speed changed to 40 Gbps
2024 Jun 25 15:48:38 fuscoastd0ca %ETHPORT-5-IF_DUPLEX: Interface Ethernet2/12, operational duplex mode changed to Full
2024 Jun 25 15:48:38 fuscoastd0ca %ETHPORT-5-IF_RX_FLOW_CONTROL: Interface Ethernet2/12, operational Receive Flow Control state changed to off
2024 Jun 25 15:48:38 fuscoastd0ca %ETHPORT-5-IF_TX_FLOW_CONTROL: Interface Ethernet2/12, operational Transmit Flow Control state changed to off
2024 Jun 25 15:48:38 fuscoastd0ca %ETHPORT-5-IF_UP: Interface Ethernet2/12 is up in mode access
As for whether or not OP has a counterfeit, or if it even matters, there’s really nothing you can do but load test. Always load test components of any kind before production.
4 Likes
Never used DAC or AOC. 10Gtek is fine as i used it for one of my friend’s project.
Vendor lock is shit. But this is great for OEM partners as they can have huge profit margins and instead of unlocking the firmware, one would prefer to have the right module, plug n play.
Quite true. A lot of them comes from China i guess.
Dang shit. I’m a Mikrotik user. Have been running Base-T for now. But the whole network will be upgraded to Fiber by next week. I just hook NIC to switch on my old Mikrotik CRS312 and the temp immediately rised from 29c to 44c at no load. I’m not sure what kind of temps would it show during the load/transfers.
Due to overheat?
Oh yes. That’s what i’m going to do
1 Like
Mikrotik documentation does state that one should not use more than two Base-T transceivers in the CRS305, and not in contiguous slots. There is a power and thermal budget—which is fair enough for a passively cooled switch.
When it comes to Mikrotik, on the one hand they at least document how many transceivers of which type can be used per switch, on the other hand, they usually omit this info from individual equipment wikis. So yes, the info is available, but it’s a bit too obscure for my taste to get to.
If you look through the table, for the CRS305, footnote 7 applies to the S-RJ01, supporting up to two of them, ditto the S+RJ10. Most vendors have some sort of limitation on copper transceivers due to heat and power of copper vs. optical modules. This is especially true for passively-cooled gear.