Overview

IEEE 802.3az Energy Efficient Ethernet (EEE) dynamically reduces port power consumption during low-traffic periods. While critical for green networking, EEE introduces unique integration and troubleshooting challenges for B2B deployments—especially with optical transceivers, stacking interfaces, and redundant power systems. This FAQ addresses both pre-sales specifications and post-sales CLI-level fixes, engineered for network architects and procurement leads.

Frequently Asked Questions

Q1: What is the maximum switching throughput of IEEE 802.3az EEE without disabling low-power idle modes?

The maximum switching throughput of IEEE 802.3az EEE is line-rate up to 100 Gbps per port, provided the link utilization stays above 30%—below which Low Power Idle (LPI) mode activates. For backplane capacity, typical enterprise switches supporting EEE offer 1.28 Tbps to 3.2 Tbps non-blocking fabric. However, when LPI wakes from sleep, the first 128–256 bytes experience a 3–5 µs wake-up penalty, which can reduce effective throughput for latency-sensitive traffic. Use the command show interface eee statistics to monitor wake events per port.

Q2: Which optical modules (SFP, SFP+, QSFP28) are fully compatible with IEEE 802.3az EEE?

Direct compatibility is limited: IEEE 802.3az EEE is natively defined for twisted-pair copper (10GBASE-T) and backplane interfaces only. For fiber-optic modules (SFP+, QSFP28), EEE support requires explicit vendor extensions—most generic optics do not support LPI. Approved compatible modules include Cisco SFP-10G-T-X (copper SFP+), Finisar FCLF-8521-3 (short-reach with EEE extensions), and Intel E40GQSFPSR (QSFP+ with low-power idle). For third-party optics, you must manually disable EEE on fiber ports (no energy-efficient-ethernet) to prevent cyclic link flaps.

Q3: How does IEEE 802.3az EEE affect stacking interface performance and virtual chassis formation?

EEE can cause stacking link flaps and split-brain scenarios. The stacking interface (typically 40G or 100G backplane) must have EEE permanently disabled because LPI sleep/wake cycles desynchronize heartbeat frames. If you observe stacking errors (e.g., "Master election loop" or "Neighbor loss"), verify with show stack-port eee status and disable EEE on all stacking ports using set stack-port eee-disable. For virtual chassis systems, ensure that only data plane ports use EEE; control plane and stacking ports must run at full power 24/7.

Q4: What are the basic CLI commands to debug a redundant PSU failover event on an EEE-enabled switch?

First, confirm PSU status and EEE power draw correlation: show power-supply redundancy and show energy-efficient-ethernet power-savings. For troubleshooting a failover where the secondary PSU fails to take over, check that EEE low-power mode did not throttle the primary PSU below the switch's minimum operating wattage. Use test eee psu-failover to simulate. A common fix: set energy-efficient-ethernet psu-min-power 80 (where 80 = 80% capacity reserve). Always collect show logging | include EEE|PSU before replacing hardware.

Q5: What is the difference in buffer-to-buffer flow control between IEEE 802.3az EEE and the previous 802.3x standard?

IEEE 802.3az EEEE introduces pause frame absorption during LPI wake-up, whereas legacy 802.3x had no power state awareness. Under 802.3x, a switch sends pause frames immediately when buffer thresholds exceed XOFF. Under EEE, the MAC may enter LPI for 3–20 µs; during this window, incoming pause frames are ignored—leading to buffer overruns. To mitigate, configure eee wake-pause-holdoff 5 (microseconds) and increase buffer reserve by 15% on EEE ports using qos eee-buffer-threshold raise 15. For zero packet loss, disable EEE on storage or SAN ports.

Q6: How do I verify that IEEE 802.3az EEE is not violating my Layer 3 OSPF adjacency timers?

EEE does not directly affect OSPF, but LPI wake delay can cause hello packet jitter. Use show ip ospf neighbor to confirm adjacency stays full; if you see flapping, run debug eee ospf-packet-timing. The safe threshold: OSPF dead interval must be > (6 x hello interval) + (4 x max EEE wake time). For a 10-second hello, set dead interval to at least 32 seconds (ip ospf dead-interval 32) on EEE-enabled ports. Alternatively, apply no energy-efficient-ethernet on specific routed interfaces where sub-50ms convergence is required.

Q7: What are the known interoperability issues between IEEE 802.3az EEE and third-party SFP+ modules from FS.com or 10Gtek?

Third-party SFP+ modules (FS.com P/N: SFP-10GSR-85, 10Gtek AXS85-192-M3) frequently misinterpret EEE LPI signals, causing excessive link up/down cycling every 100–200 ms. Symptoms include high CRC error rates and autonegotiation failures. Resolution: on each interface, run default eee-capability to disable EEE advertisement, then force speed 10000 full-duplex and no negotiation auto. For mixed fleets, create a port profile: interface range gigabitethernet 1/0/1-24\n eee-disable\n auto-negotiate disable\n speed 10000. Always test with show interface counters eee-errors.

Q8: After a power brownout, my EEE switch fails to bring up redundant PSU—is this an EEE firmware bug?

Yes, this is a known firmware race condition affecting certain Broadcom Trident 3-based switches when EEE was active during brownout. The EEE power-state machine incorrectly flags the redundant PSU as "low power capable" and disables its 12V rail. Fix: upgrade to firmware version 8.4.2+ where eee-psu-recovery is patched. Workaround without reboot: force PSU re-init via power-supply redundancy force-switch 2 then clear eee power-state volatile. To prevent recurrence, set energy-efficient-ethernet psu-failover-mode aggressive.