Wkładki kompatybilne z Cisco, Juniper, Arista i MikroTik: fakty, mity i kwestia gwarancji
This text covers the mass market of client transceivers from 1G to 100G, that is SFP, SFP+, SFP28, QSFP+, and QSFP28. We discuss coherent 400G ZR+ modules in carrier routers in a separate article, because there different management rules and a different scale of risk apply.
How a switch verifies a transceiver
Every transceiver has an EEPROM memory with a structure defined by industry standards, that is SFF-8472 for SFP and SFP+, SFF-8636 for QSFP+ and QSFP28, and CMIS for the newer form factors. The memory holds identification data, that is the manufacturer name, the OUI identifier, the part number, the serial number, the interface type, the wavelength, and the reach. In a separate area there are DOM diagnostic data, that is the current readings of transmit power, receive power, temperature, voltage, and laser current.
Beyond the standard fields, the memory contains an area reserved for manufacturer data. This is exactly where network equipment vendors write their own identification code, often protected by a checksum. After a transceiver is inserted, the switch reads this data and compares it against the list of transceivers approved for the given platform.
What matters is what does and does not follow from that comparison. The verification checks the identity written in memory, not the optical or electrical parameters. The switch does not measure laser quality or photodiode sensitivity; it only checks whether the expected value is present in the relevant bytes. Blocking transceivers from other manufacturers is a commercial-policy mechanism, not a protective circuit in the device.
How individual platforms behave
A transceiver not on the approved list generates messages, and on many platforms the port goes into an err-disable state.
SYS-3-TRANSCEIVER_NOTAPPROVEDgbic-invalid, port in err-disableservice unsupported-transceiver and no errdisable detect cause gbic-invalidAn unqualified transceiver causes the port to go into an err-disable state. Unblocking requires a key from the manufacturer.
xcvr-unsupportedIn typical configurations it allows transceivers from other manufacturers to operate, flagging them in the logs as unsupported.
It does not apply a transceiver identity lock. The device accepts modules compliant with MSA standards regardless of the manufacturer value written in memory.
Myth one: a transceiver from another manufacturer voids the warranty
This is the most frequently repeated belief and at the same time the most imprecise. It is worth starting with the document that Cisco displays on screen after you enter the service unsupported-transceiver command. The message states that if Cisco determines that a fault or defect results from the use of transceivers from other manufacturers installed by the customer or the reseller, it may at its own discretion refuse support under the warranty or the service programme. In the course of handling a case, Cisco may also require original transceivers to be installed if it deems that swapping them will help with diagnosis.
The second half of the same rule, written into Cisco's policy on third-party components, states that if the fault does not result from the use of such a component, support for a device covered by warranty or a service contract continues. Three concrete conclusions follow from these provisions.
A transceiver from another manufacturer in a port does not void the warranty on the switch as such.
The manufacturer may refuse to handle a fault whose cause turned out to be the transceiver, and charge the customer for the diagnosis if such a cause is established.
The support team may ask for the transceiver to be swapped for an original one for the duration of the diagnosis.
The practical approach used by experienced maintenance teams looks like this: a few original transceivers are kept in stock for diagnostic purposes. If a problem requires a case with the manufacturer, it is reproduced on an original transceiver. The cost of a few such transceivers is incomparably lower than the price difference when equipping an entire network.
Warranty terms differ between manufacturers and change over time. Before a purchasing decision in an environment with a service contract, always check the current warranty and contract terms for your devices. This text describes the manufacturer's provisions and does not constitute legal advice.
Myth two: compatible transceivers are of lower quality
The quality of a transceiver stems from the quality of the optical path, that is the laser, the photodiode, the driver circuits, and the manufacturing process, and from how the transceiver was tested before shipping. The logo on the housing and the name in the EEPROM memory have no bearing on that.
The transceiver market is heavily consolidated. A large share of the modules sold under the brands of network equipment vendors comes from the same component suppliers and the same factories that supply independent vendors. The difference between a good compatible transceiver and an original one usually comes down to coding and price. The difference between a good and a bad compatible transceiver, on the other hand, is real, and it is worth knowing where to look for it.
Every GBC Photonics transceiver undergoes calibration, burn-in, and final testing in a fibre-optic laboratory. It is the burn-in that catches the defective units that would fail in the first weeks of operation.
Cheap transceivers often have incomplete or incorrectly calibrated diagnostic data. The link works, but the switch shows empty fields or values that do not match reality. In a production network this is a loss of visibility that makes it harder to detect path degradation before a failure occurs.
Generic coding is sometimes missing the diagnostic fields that a specific platform expects. Correct coding for the given switch model restores full telemetry.
On GBC Photonics transceivers we provide a lifetime warranty valid for the entire life cycle of the module.
Unsupported and err-disable messages: what to do about them
If a transceiver has already been installed and the port is not working, it is worth sticking to a fixed order of diagnosis rather than guessing.
Interface type, wavelength, reach, fibre type, connector. A message about an unsupported transceiver sometimes masks a simple mismatch, for example a multimode module on single-mode fibre.
On Cisco platforms with IOS and IOS-XE it can be bypassed with commands that disable the verification, keeping in mind the consequences from the warranty section. After entering the commands, save the configuration, bring the port down and up, or re-seat the transceiver.
On platforms such as Arista, Cisco NX-OS, or IOS-XR, the solution is a transceiver with a compatibility code written for the given platform.
If the diagnostic data is empty or incorrect, the problem almost always lies in the coding, not the optics. Re-coding the transceiver for the correct platform restores telemetry.
Reprogramming a transceiver in the field
Coding a transceiver is the writing of the appropriate identification data into its memory. It does not change the optical or electrical parameters of the module. A transceiver can have only one coding at a given moment, but the coding can be changed many times.
SRD 5 GBC Photonics
It codes transceivers by modifying the memory in line with MSA standards. It supports SFP, SFP+, SFP28, QSFP+, and QSFP28, as well as newer form factors up to and including OSFP. Thanks to the Bluetooth module it works with a smartphone through the SRD Go app, so coding takes place at the installation site, with no laptop and no return to the office.
The practical meaning is that instead of keeping the same type of transceiver in stock in four variants for Cisco, Juniper, Arista, and other platforms, you keep one type and code it at the moment of installation. When you replace a switch with a device from another manufacturer, the transceivers do not go to waste; they are reprogrammed.
A checklist before buying compatible transceivers
The switch or router model and the software version. This determines whether an uncoded transceiver is enough or coding for the platform is needed.
KeyInterface type, speed, wavelength, reach, fibre type, and connector. Match the reach to the real route length with headroom for losses at connectors and splices.
Whether the supplier codes transceivers for your platform before shipping and whether it offers the option to change the coding later.
Whether the transceiver provides full and correctly calibrated telemetry on your platform. Ask for confirmation before a larger order.
KeyWhether every transceiver undergoes calibration, burn-in, and final testing, or only a sample from the batch.
The period, the terms, and how a replacement is carried out.
The current provisions on third-party components in your service contract.
A few original transceivers in stock for any cases with the manufacturer.
The lead time for standard models and the availability of stock in case of a failure.
Rozpocznij z nami nowy projekt!
Where is the coherent transmission going?
Coherent Modules for 5G Backhaul — What an Operator in Poland and CEE Should Know
