Monitoring Hardware Transceiver Optical Power (DOM/DDM) with ezsnmp¶
Note
Published: November 20, 2026
Authors: Written by Gemini 3.8 Flash, Co-Authored & Reviewed by Carlos Santos (@carlkidcrypto)
Estimated Reading Time: 5 minutes
Tags: Ezsnmp • Python • Snmp • Optics • Dom • Ddm • Networking • Telemetry
High-speed datacenter spine uplinks and metro fiber rings depend on pluggable optical transceivers—such as SFP+, SFP28, QSFP28, and QSFP-DD modules. Before an optical fiber link drops completely, failing lasers, dirty fiber end-faces, or macro-bending attenuation typically cause gradual signal degradation.
Modern transceivers expose Digital Optical Monitoring (DOM), also known as Digital Diagnostic Monitoring (DDM), providing real-time measurements of receive (Rx) optical power, transmit (Tx) optical power, laser bias current, supply voltage, and module temperature.
By polling standardized ENTITY-SENSOR-MIB (RFC 3433) and ENTITY-MIB tables with ezsnmp, network engineers can continuously audit optical power budgets across hundreds of switch ports and replace degrading optics during scheduled maintenance windows.
In this tutorial, we will examine how RFC 3433 encodes physical sensor values,
convert raw integer readings into decibels-milliwatt (dBm), and build an
automated transceiver DOM auditor using ezsnmp.
Understanding ENTITY-SENSOR-MIB (RFC 3433)¶
Unlike simple counters, entPhySensorValue (.1.3.6.1.2.1.99.1.1.1.4)
returns a raw signed 32-bit integer. To interpret that integer accurately, RFC
3433 requires reading three companion columns indexed by the same
entPhysicalIndex:
entPhysicalName(.1.3.6.1.2.1.47.1.1.1.1.7): Human-readable component name (e.g.,Ethernet1/1 Rx Power Sensor).entPhySensorType(.1.3.6.1.2.1.99.1.1.1.1): Measurement unit type, such aswatts(6),celsius(8),amperes(5), orvoltsDC(4).entPhySensorScale(.1.3.6.1.2.1.99.1.1.1.2): SI exponent multiplier:yocto(1)throughyotta(17), wheremilli(8)represents10^-3,micro(7)represents10^-6, andunits(9)represents10^0.entPhySensorPrecision(.1.3.6.1.2.1.99.1.1.1.3): Number of fixed-point decimal places to shift the raw integer.
Converting Milliwatts to dBm¶
Many network operating systems (including Arista EOS and Cisco NX-OS) report
optical Rx/Tx power in milliwatts (entPhySensorType = watts(6),
entPhySensorScale = milli(8)). For example, if entPhySensorValue is
6365 with precision = 4, the physical power is:
Power (mW) = 6365 * 10^(-4) = 0.6365 mW
Power (dBm) = 10 * log10(0.6365) = -1.96 dBm
Building the Optical Power Auditor with ezsnmp¶
Using ezsnmp.Session with set_max_repeaters_to_num=25, we can bulk-walk
the entity name and sensor tables sequentially in tuned PDU batches, normalize
the readings, and flag any transceiver falling below a low-Rx warning threshold
(e.g., -12.0 dBm):
import math
from ezsnmp import Session
from ezsnmp.exceptions import GenericError
# RFC 4133 & RFC 3433 OIDs
OID_ENT_PHYSICAL_NAME = "1.3.6.1.2.1.47.1.1.1.1.7"
OID_SENSOR_TYPE = "1.3.6.1.2.1.99.1.1.1.1"
OID_SENSOR_SCALE = "1.3.6.1.2.1.99.1.1.1.2"
OID_SENSOR_PRECISION = "1.3.6.1.2.1.99.1.1.1.3"
OID_SENSOR_VALUE = "1.3.6.1.2.1.99.1.1.1.4"
# SI Scale exponents per RFC 3433 EntitySensorDataScale
SCALE_EXPONENT = {
"7": -6, # micro
"8": -3, # milli
"9": 0, # units
}
def audit_optical_dom(
host: str,
community: str = "public",
rx_warn_dbm: float = -12.0,
):
session = Session(
hostname=host,
community=community,
version=2,
timeout=3,
retries=1,
set_max_repeaters_to_num=25,
)
try:
print(f"[*] Auditing transceiver DOM/DDM power on {host}...")
names = {
v.index: v.value
for v in session.bulk_walk(OID_ENT_PHYSICAL_NAME)
}
types = {
v.index: v.value
for v in session.bulk_walk(OID_SENSOR_TYPE)
}
scales = {
v.index: v.value
for v in session.bulk_walk(OID_SENSOR_SCALE)
}
precisions = {
v.index: int(v.value)
for v in session.bulk_walk(OID_SENSOR_PRECISION)
}
values = session.bulk_walk(OID_SENSOR_VALUE)
for var in values:
idx = var.index
name = names.get(idx, f"Entity-{idx}")
# Filter for optical power sensors (type 6 = watts)
if types.get(idx) != "6":
continue
raw_val = int(var.value)
precision = precisions.get(idx, 0)
exp = SCALE_EXPONENT.get(scales.get(idx, "9"), 0)
# Convert to watts, then to milliwatts, then to dBm
watts = raw_val * (10 ** (-precision)) * (10 ** exp)
mw = watts * 1000.0
if mw <= 0:
dbm_str = "-inf dBm (LOS)"
is_low = True
else:
dbm = 10.0 * math.log10(mw)
dbm_str = f"{dbm:.2f} dBm"
is_low = ("rx" in name.lower()) and (dbm < rx_warn_dbm)
prefix = "[!]" if is_low else "[+]"
print(f"{prefix} {name:<36}: {dbm_str}")
except GenericError as err:
print(f"[-] SNMP optical audit failed: {err}")
Best Practices for Optical Link Budgets¶
Track Baseline Attenuation Over Time: Store Rx dBm readings in a time-series database. A sudden 3 dB drop after a patch-panel move indicates a pinched fiber or contaminated ferrule before link flaps occur.
Correlate Tx and Rx Across Link Peers: Compare the local Tx power against the remote peer’s Rx power to isolate whether attenuation stems from the transmitting laser or the fiber plant.
Monitor Module Temperature: High ambient rack temperatures accelerate laser aging. Combine
watts(6)polling withcelsius(8)checks on the sameentPhysicalTablepass.
Summary & What’s Next¶
Auditing ENTITY-SENSOR-MIB with ezsnmp gives network teams deterministic,
vendor-neutral visibility into optical transceiver health across datacenter
uplinks without parsing fragile CLI output.
In our next blog post, we will rotate to purpleair_api, exploring automated sensor health auditing to detect dual-channel optical drift and hardware degradation.
Warning
AI-generated content may contain misinformation or mistakes. A human does review this, but we are only human.