IT Problème ... Magic Cable

Solution du IT Problème "ping impossible".

Comme l'a dit Christophe en commentaire, il suffisait tout simplement de faire un ping en explicitant le nom de la VRF sur R1 : ping vrf CUST 192.168.0.1.

En effet, l'interface est configurée en étant membre de la VRF CUST. Donc elle n'est plus dans le contexte global du routeur. Il est d'ailleurs possible de vérifier cela en tapant un show ip route sur R1, on voit que le réseau n'est plus sur le routeur (ce que j'appelle contexte global), par contre un show ip route vrf CUST permet de voir l'existence de la route sur la vrf CUST.

J'en profite aussi pour dire que la solution Cisco EVN (Easy Virtual Network) facilite la gestion des VRFs. En effet, il suffira alors de se mettre dans la VRF CUST (via routing-context vrf CUST), puis de taper ping 192.168.0.1 pour faire un ping dans la VRF CUST.

A new era with ... OpenFlow?

I wanted to write a post about OpenFlow, but it's holidays, and you know... :)

So the OpenFlow protocol is related to the the "Software-Defined Networking" (or SDN). The SDN is the new paradigm where the control plane and the data plane are distinct. The hardware is used to transmit packets and the processing tasks is done on more powerful equipments.

And here comes the OpenFlow protocol. The OpenFlow protocol is the protocol used between the network equipment and the processing equipment (the OpenFlow controller).

http://www.openflow.org/documents/openflow-wp-latest.pdf

So how all of this is supposed to work?

In fact, the network equipment will send all the frames / packets received to the controller for processing through a secure channel. The controller will reply to the network equipment with the output interface on which the packet should be transmit. The equipments will only accomplish switching operation.

What are the advantages?

• Test of new protocols on the network: the OpenFlow protocol has been designed by academics for research purposes. [1]
• Manufacturer independent: if the equipment support OpenFlow, it can be used in the network with others OpenFlow enabled equipments
• Centralized management : all the equipments are connected to the OpenFlow controller.

OpenFlow and the SDN are opening a new era: manufacturers will deliver equipments with high line card and the packet processing will be done at the controller. 

OpenFlow is still under development, but Google has already deployed OpenFlow on its production network [2] [3].

A little video where Urs Hoezle is talking at the Open Networking Summit 2012.

And Cisco is releasing equipments which can support OpenFlow, it's the "Cisco Open Network Environnement".

In this hypothetical Cisco Open Network Environnement, the OpenFlow controller could become a centralized station for all the different services, the Cisco Prime Unified Security Open Orchestrator (or Cisco PUSOO), from this controller: centralization of energy management (we'll suppose that manufacturers will release EMAN WG MIBs), centralization of security policy, centralization of network management and monitoring, all in one. Of course, there is currently no Cisco Prime Unified Secure Open Orchestrator or maybe there is one at the beta version.

Let's imagine a bit further ... in a future, where all network equipments will request to a global controller, and this global controller is running with a AI, it will be Skynet (the famous AI from Terminator).

With OpenFlow, SDN and the cloud, maybe the beginning of the RaaS (Routing as a Service)? 

References

[1] McKeown N., Anderson T. et al. OpenFlow : Enabling Innovation in Campus Networks, March 2008

[2] Google Inter-Datacenter WAN with centralized TE using SDN and OpenFlow, August 2012

[3] Merritt R., Google describes its OpenFlow network, EE Times, April 17th 2012

IT Problème ... ping impossible

Solution du IT Problème "Fibre Optique".

Théoriquement, l'atténuation sur la fibre devrait être d'environ 0,1 dB (il y a environ 500m de fibre, l'atténuation théorique pour une longueur d'onde de 1310 nm est de 0,2 dB/km).

Or ce n'est pas le cas ...
Il y a un raccord de 2 fibres (entre segment 1 et 2) au niveau A, cela provoque une atténuation de l'ordre de 0,02 dB.
Il y a un connecteur de fibres (entre segment 2 et segment 3) au niveau B, cela entraine une atténuation de l'odre de 0,2 dB.

Interprétation de l'OTDR

Voila pour l'explication du problème précedent, maintenant un nouveau problème plus simple.

SPF incrémentiel

SPF ou l'algorithme de Dijkstra permet aux protocoles de routage à états de lien d'avoir une table de routage à partir de la table de topologie.

L'inconvénient est qu'il faut recalculer la table de routage avec SPF après chaque LSA.

Incremential SPF

SPF or Dijkstra algorithm gives the routing table to link-state routing protocols from the topological table.

The drawback is that the routing table must be recomputed with SPF after every LSA.