EXOS VLAN mismatch detection

[M_Nees]

Configuring VLANs manually on switch ports is still a daily business for all network admins.

Support by OS to detect errors is very helpful.

EXOS can detect VLAN mismatch configuration (with EDP - Extreme Discovery Protocol) but with some bad limits:

• till 30.3 EXOS can only detection which VLANs are exists on the neighbor switch (no information about tag or untagged)
• since EXOS 30.3 it can also show which VLAN is native (UT) and which are tagged (better)
• it can detect on a neighbor link (ISL) that are different untagged VLANs are mapped and can generate an alarm (because in most cases this is an error). But an error message occurs only one time during the switch is running, no further alarm will generated

As we know that from other vendors this can and needed to be solved in a better way.

Therefore we need enhancement that EXOS vlan mismatch detection can do this:

• detect not only native vlan mismatch, detect also tagged VLAN mismatch on a ISL
• generate Alarms if this is detected
• Alarms needed as syslog entry in switch Log and also as SNMP trap to XMC
• Alarm rate have to be customizeable (not only one during switch booted up)
  rate should configureable - one alarm event per minute / per hour / per day
• during configuration state alarming should have to be disabled globally (easily)
• Alarming have to be enabled/disabled on a per port base:
  Alarming on Uplink is very interesting otherwise on a special ports, where VLAN translation is configured, Alarming have to be turned off

BR,

Matthias

[mitchjreyes]

Use Orchestration mode to sync both ends of the cluster.

[M_Nees]

Hi Alex,

right now i used the mentioned orchestration mode on 2 mlag EXOS devices.

Configuring the next 2 corresponding VOSS Switches there is sadly no orchestration mode ;-(

We should convince that VOSS also implement such an orchestartion mode !

BR,

Matthias

[AlexN]

Yes, FA on EXOS definitely needs enhancements, and we are working on them. 30sec timer is one of them :), it should react realtime, not but by LLDP default keep alive intervals
However I referred to all fabrics, not just SPBm, in EXOS to EXOS that would be BGP EVPN auto peering..

[M_Nees]

Hi Alex,

thanks for your reply.

Let’s answer your questions.

1. Uplink - why not everywhere Fabric Attach?
   + Not in every customer setup EXOS Switches are attached to a VSP Switch!
   Think about a lot of ISL between 2 EXOS Access Switches which are manually configured. Altough EXOS support MVRP, there are always manually ISL config necessary and used!
   + I like Fabric and i like FA too, but also FA have some disadvantages. FA depends on slow LLDP signaling interval (which is default 30sec) so sometimes it takes a lot of time till VLANs are requested from VSP. i hope this will be changed to event triggered signaling that EXOS can request needed VLANs at once, not waiting on a LLDP interval. 
   + Not all EXOS customers like VSP CLI! So some customers prefer similar OS means similar CLI commands  from access to core switch.
   + I love automation and use it where ever it is possible, so Fabric Attach or MVRP is a good choice. And helps avoiding config errors.
    
2. i know orchestration mode - it helps avoiding errors. i think my request and orchestration giving the customer the needed tools avoiding config errors. so i think both tools are fitting good together. 

BR,

Matthias