Category: Ethernet

STP Quick Reference

  1. Port Role:
    1. Designated:
      1. one per segment
      2. it is the port closest to the root bridge.
      3. it sends BPDU to the segment to the root or alternate ports.
      4. all ports on the root bridge are in designated role.
      5. It is in forwarding state.
    2. Root:
      1. one per switch, except the root bridge itself.
      2. it is the port that leads to the root bridge.
      3. it receives BPDU from the upstream switch.
      4. It is in forwarding state.
    3. Alternate:
      1. Alternate ports to the root bridge, but are not the root port.
      2. it receives BPDU from upstream switch.
      3. It is in blocking state.
  2. STP Timers:
    1. Max Age:
      1. It is how long a switch would store BPDUs it received on all its ports.
      2. When a switch didn’t receive BPDU on its port up to the Max Age, it will send its BPDU out to the port instead.
      3. Alternate port in blocking mode wait for Max Age and 2 times forward_delay timer to expire before transition into forwarding state
      4. Max Age default is 20 seconds
    2. Delay Timer:
  3. LoopGuard:
    1. When connection to the root bridge is broken, Alternate Port stopped receiving the root BPDU, and will transition to Designated Port and Start Forwarding traffic after Max Age + 2x Delay.
    2. However, sometimes that behavior could create a loop. when there is unidirectional link issue, Alternate Port also did not get BPDU from its upstream designated port. When the Alternate Port starts transitioned to Designated Port and start forwarding traffic, that will create a unidirectional loop in the switch network, in the reverse direction of the failure, because the upstream switch port still can accept packet in its receiving direction.
    3. The is a architecture issue with STP because STP fail open instead of fail close.
    4. LoopGuard tells Alternative Port not to change to forward state until it hears any BPDU from its upstream switch’s designated port.
    5. You want to configure LoopGuard on all Root and Alternate Ports
  4. Backbone Fast:
    1. All a indirect link failure, an Alternate Port could take as long as 50 seconds (Max Age + 2x Delay) before it transitions to forwarding state.
    2. Backbone Fast implements a layer 2 PING liked mechanism. Once enable, switches can send query to the root bridge and root bridge would response.
    3. The switch timeout its Max Age immediately after it receives the response from the root bridge.
    4. The reduce the potential convergence time from 50 seconds to 30 seconds.
    5. Backbone Fast is native to RSTP. So you do not need to enable it manually when running RSTP.
  5. Uplinks Fast:
    1. On classic STP, when a directly connected uplink failed, the switch alternate port would wait max 2x delay timer because it transition to root or designated port and start forwarding traffic. So the max convergence time is 30 seconds.
    2. The convergence time is 2x of the delay timer because an blocking port needs to go through listening state (one delay timer) plus learning state (another delay timer) because it is fully transitioned into forwarding state.
    3. Once enable, enable experience uplink failure on the switch root port, the switch immediately transition the Alternate Port with the lower cost to the root bridge into forwarding state, thus reduce the convergence time from 30 second to less than 1 second.
Advertisements

IEEE 802.1ag and ITU-T Y.1731

Overview

  • IEEE 802.1ag and ITU-Y Y.1731 are pretty much the same protocol. It is the OAM protocol for Carrier Ethernet

External Links:

EoMPLS

  1. Ethernet over MPLS is a point to point technology. Use VPLS for multiple access Ethernet over MPLS
  2. EoMPLS has two modes:
    1. Ethernet Port mode
      1. Transports whole ethernet port, the ethernet could have no vlan tag or could have vlan tag.
      2. The port can be trunked, which let it carry multiple vlans over one EoMPLS tunnel
    2. Ethernet VLAN mode
      1. Is used to transport only one VLAN per pseudo wire over an EoMPLS tunnel
      2. The advantage is the ability to terminate VLANs on different remote PEs

QinQ

QinQ is simply an Ethernet protocol that allows two VLAN tags on Ethernet frames.

A VLAN tags is 4-byte long. QinQ allows two VLAN tags, a total of 8-byte on Ethernet frames.

The application in for Service Provider to have the ability to carry multiple customers VLAN (C-VLAN) within their own VLAN (S-VLAN). Commonly used in MetroEthernet networks.

QinQ is Cisco proprietary, the IEEE standard is 802.1ad. Both are essentially the same, except that 802.1ad allows more than two tags.

Eternal Links:

http://en.wikipedia.org/wiki/IEEE_802.1ad

http://www.juniper.net/techpubs/en_US/junos12.3/topics/concept/qinq-tunneling-qfx-series.html