Optiken

DOM status critical or warning: what your switch is telling you

Your dashboard shows DOM critical or a laser bias current high alarm. What the four thresholds mean, who sets them and the order to fix it in.

5 Min. LesezeitVon Yvan Watchman

Your Meraki dashboard turns a port red with "DOM status critical", or Junos reports "Laser bias current high alarm: On". Is the optic dead, or is there dust on a connector? This post explains what the thresholds mean and who sets them, what each alarm usually points to, and the order to work through it, cheapest step first.

Critical or warning: four limits per value

DOM/DDM measures five values: temperature, voltage, laser bias current, Tx power and Rx power. Under SFF-8472, each has four thresholds: high alarm, high warning, low warning and low alarm.

A warning means you are outside the normally guaranteed bounds, but the link doesn't have to fail yet. An alarm is likely tied to a link that no longer works properly and calls for immediate action. Meraki shows high and low alarms as Critical and high and low warnings as Warning. Arista puts ALARM or WARN in the Indicator column; Cisco IOS XE flags the value with ++, +, - or --.

Who sets the thresholds, and why your third-party optic complains

Your switch doesn't set those limits. The transceiver manufacturer does. SFF-8472 calls them "factory preset": they vary by technology and implementation, and a manufacturer may adjust them in a vendor-specific way, for temperature for example.

That is where odd warnings come from. Two optics of the same type can carry different limits. Some modules are also "externally calibrated": they report raw A/D counts the switch has to convert, and if it compares after converting, it must convert the thresholds the same way. Get that wrong and you can see a warning on a link that runs fine. Meraki is upfront about it: it only guarantees the accuracy of its own SFPs.

And DOM is not a test instrument. SFF-8472 requires ±3 dB accuracy for Tx and Rx power and ±10% for bias. A single reading that ticks just past a limit tells you little. A change against the same module a month ago tells you a lot more. Record a baseline per module and watch the trend.

What each alarm usually means

  • Rx power low — too little light arrives: dirt, damaged fiber, too much loss on the path, or the transmitter on the far end. Cisco points to the far side and the cable first.
  • Rx power high — too much light, typically a long-reach optic on a short link. Cisco specifies a 5 dB attenuator for its 10G ER below 20 km.
  • Tx power low — your own transmitter is putting out less, so look at the local optic. SFF-8472 says Tx data is not valid while the transmitter is disabled, so check the port is actually up first.
  • Laser bias current high — the laser needs more current for the same light. SFF-8472 names this as a possible end-of-life indicator, but bias also rises with temperature. An Intel patent notes that a fixed bias limit can raise a warning for exactly that reason, with the laser nowhere near failure.
  • Laser bias current low — the laser is barely drawing current. Combined with low Tx power, that points at the optic itself.
  • Temperature high — poor airflow, or a module rated for 0 to 70 °C in a cabinet that runs too hot. MikroTik shuts an SFP down at 95 °C by default.
  • Voltage — move the module to another port. If the alert follows it, it's the optic; if it stays with the port, look at the switch.

Work through it in this order

Microsoft monitored 350,000 links across 15 data centers and traced the cause of packet corruption in over 300 tickets. Contaminated connectors accounted for 17 to 57%, damaged or tightly bent fiber for 14 to 48%, and bad or loosely seated transceivers for 6 to 45%. Decaying lasers: under 1%. When technicians followed advice based on Tx and Rx from both ends, 80% of repairs worked first time, up from 50%. Swapping the optic feels like the quickest fix, but on links with packet errors the problem sits in the glass more often than in the optic.

  • Read DOM on both ends of the link, including the thresholds.
  • Compare against the baseline for the same module, not just against the limits.
  • Check module temperature and airflow first for any bias or temperature alert.
  • Inspect and clean every connector in the path, then measure again.
  • Check the patch cord for kinks and swap it for a known-good one.
  • Reseat the transceiver.
  • Replace the transceiver if Tx and bias on the same side stay off.

To read the values, use show commands that change nothing: show interfaces Te1/0/1 transceiver detail (Cisco IOS XE), show interfaces diagnostics optics xe-0/0/0 (Junos), show interfaces Ethernet1 transceiver dom thresholds (Arista) and /interface ethernet monitor sfp-sfpplus1 (MikroTik, which shows no thresholds: take those from the datasheet). Meraki shows everything on the Ports tab, with history.

When it really is the optic

Low Tx power on the transmitting side, with low Rx on the far end, after cleaning and a fresh patch cord: that is the optic. Same if bias creeps up over weeks while temperature stays flat, or if the alert follows the module to another port. No spare on the shelf? Read how to sort spares and same-day replacement.

Clean with a Clear Connection Tool, replacement optics ship same day in Amsterdam, and link testing is available through our services.

Rather not do this by hand every time? That's why we're building Praevisio, currently in beta with a waitlist. Praevisio tracks each module's DOM values against its own baseline, alongside CRC and FEC errors and flapping links, and advises the cheapest cause first. For example: "Clean the connection first. Praevisio then measures again for fifteen minutes. If the advice is still the same, then: swap the transceiver." Your data stays in the Netherlands.

Yvan Watchman

Über den Autor

Yvan Watchman

Gründer von FiberNxt und Xyphen IT

Yvan Watchman ist Gründer von FiberNxt und Xyphen IT. Er baut und betreibt seit Jahren Netzwerke und Rechenzentrumsinfrastruktur in und um Amsterdam.

LinkedIn-Profil

Passende Produkte und Dienstleistungen

Häufig gestellte Fragen

What is the difference between a DOM warning and a DOM alarm?

According to SFF-8472, a warning means a value is outside the normally guaranteed bounds, without the link necessarily failing. An alarm is likely tied to a link that no longer works properly and calls for immediate action. Meraki shows alarms as Critical and warnings as Warning.

Who sets the DOM thresholds on my SFP?

The transceiver manufacturer. The thresholds live in the module's EEPROM and differ per type and per brand. Your switch only reads them and compares the live value against them.

Should I replace an SFP with a laser bias current high alarm?

Not straight away. Bias also rises with temperature, so check the module temperature and airflow first. If bias climbs over weeks while Tx power holds or drops at a steady temperature, the laser is probably wearing out and you plan a replacement.

Weiterlesen