MEF: Benefits of Multiple Class Of Service Model For Mobile Backhauls

According to all wireless operators, delivering the bandwidth required in the 4G-LTE wireless backhaul is “the single biggest challenge and operating cost in the industry.” Currently a single class of service model is being deployed by the mobile operators at aggregation side where the traffic is classified and there after all the traffic is forwarded to a single EVC without defining the priority. Single-CoS scenario, however, does NOT respect these differing priorities which means at any point of time access provider cannot use the stat multiplexing.
To address the issue, a new multiple CoS model is added for the access providers. The configuration of the Mobile Operator CE devices is similar to that for the Single CoS Ethernet Mobile Backhaul alternative. The primary added configuration is ensuring that each Frame it passes to the Access provider is marked with the appropriate priority for that Frame. It does so by marking the Priority Code Point (PCP) bits in the Frame.

Mbh multi co-s positioning paper 2012011

View more documents from shivlu.

Click Here To Read Rest Of The Post...

802.3ah - Ethernet OAM - Part 1

802.3ah is a link level Metro Ethernet OAM which is used for the last mile monitoring purpose. It cannot scale more than a single hop. It can transmit at max rate of 10 frames per second which is equivalent to 100ms transmission time for single frame. At least we need 3 frames to detect the link loss with high cpu usage which will contribute almost 300ms to acheive only link level convergence time.

Functions:-
OAM discovery
Remote Failure Indication remote loopback
Port Level Loopback
Fault Isolation
Perfomrance and Status Monitoring


Click Here To Read Rest Of The Post...

G.8032 Ethernet Ring Protection Switching

Spanning tree protocol is used in layer 2 networks for logical loop avoidance because laer 2 PDU doesn’t have any TTL value defined in their header like layer 3 packets does. Convergence and scalability is the biggest challenges for the evolving Metro Ethernet networks. RPR aka Resilient Packet Ring was proposed by IEEE (802.17) with 50ms convergence with new mac 802.17 header enhancements, which was not compatible with existing Ethernet header consequences lot of new efforts required to build a new code.
So a protocol is require which could be fitted in the current Ethernet header with less efforts and provides faster convergence than the legacy protocols. ITU came up with Ethernet Ring Protection Protocol aka ERPS(G.8032) which uses the SONET/SDH APS(Automatic Protection Switching) protocol for protection switching and Ethernet operation and maintenance for detecting of faster failures which is less than 50ms.
In an Ethernet ring, without congestion, with all nodes in the idle state with less than 1'200 km of ring fibre circumference, and fewer than 16 to 255 nodes, the switch completion time (transfer time) for a failure on a ring link shall be less than 50 ms. In fact, practically it is not possible.

In summary, G.8032 ERP is developed to meet the following objectives:
•To provide efficient network connectivity.
•To provide rapid service restoration (sub-50ms).
•To support multiple E-Services (e.g., E-LINE, E-TREE, E-LAN).
•To be client and server layer agnostic. That is, G.8032 ERP can be supported over (virtually)
any physical/server layer and can transport (virtually) any Ethernet client.
•To utilize existing IEEE 802.1 bridging and IEEE 802.3 MAC hardware. Thus is simply a
software increment on existing Ethernet switching equipment.
•To support flexible deployment models. To be deployed in Access, Metro, and Core network
applications.
•To leverage Ethernet PHY bandwidth (e.g., 1/10/40/100GbE MAC), cost and time-to-market curves in support of cost effective and large bandwidth rings.
•To be standardized in ITU-T SG15/Q9.
•To translate into lower OPEX and CAPEX for the service provider (Very Important).

Click Here To Read Rest Of The Post...

MEF Service Level Aggrements In Interconnect Networks

Topic Covered:-
1. Three Class Of Service(CoS) Model Framework
2. Delivering Service Level Aggrement
3. Bandwidth Profiling Model
4. Performance

MEF Service Level Aggrement

View more presentations from shivlu.

Click Here To Read Rest Of The Post...

How does two different EVC communicate when using different encapsulation

In last post, ports are communicating with the same vlan encapsulation type. But how will the different encapsulation communicate with each other.

We know when we configure dot1q encapsulation on router’s port, it always remove the tag during ingress and add the tags during egress forwarding. If the router is configured with multiple sub-interfaces and every sub-interface is having different dot1q encapsulation; in that case the router always remove the vlan tags while ingress from one sub-interface and adds the tags during egress of that sub-interface. To perform this you must have router or layer 3 capable switch which could perform this. (Without Layer 3 device)

EVC architecture also works on the same mechanism, as per given configuration, both the service instances cannot communicate with each other as their encapsulation is different. One service instance is carrying 109 and other is carrying 110.  

Switch (config)# interface gigabitethernet0/10

Switch (config-if)# switchport mode trunk

Switch (config-if)# switchport trunk allowed vlan none

Switch (config-if)# service instance 101 Ethernet

Switch (config-if-srv)# encapsulation dot1q 109

Switch (config-if-srv)# bridge-domain 103

Switch (config)# interface gigabitethernet0/11

Switch (config-if)# switchport mode trunk

Switch (config-if)# switchport trunk allowed vlan none

Switch (config-if)# service instance 101 Ethernet

Switch (config-if-srv)# encapsulation dot1q 110

Switch (config-if-srv)# bridge-domain 103

Both the service instance can communicate if and only if they receive the same encapsulation type. To achieve this we must use the rewrite keyword in any of the service instance. Rewrite keyword performs the pop operation during ingress consequently vlan tag gets removed and symmetric keyword helps to perform the inverse function of adding vlans during egress.   

Switch (config)# interface gigabitethernet0/10

Switch (config-if)# switchport mode trunk

Switch (config-if)# switchport trunk allowed vlan none

Switch (config-if)# service instance 101 Ethernet

Switch (config-if-srv)# encapsulation dot1q 109

Switch (config-if-srv)# bridge-domain 103

Switch (config)# interface gigabitethernet0/11

Switch (config-if)# switchport mode trunk

Switch (config-if)# switchport trunk allowed vlan none

Switch (config-if)# service instance 101 Ethernet

Switch (config-if-srv)# encapsulation dot1q 110

Switch (config-if-srv)# rewrite ingress tag pop 1 symmetric

Switch (config-if-srv)# bridge-domain 103

Click Here To Read Rest Of The Post...

Metro Ethernet EVC features

When we configure service instance on any port, that port becomes a pseudo port. The service instance defined under any port has some rules to match the incoming frames from CPE. Those rules defined the Ethernet flow points. Without the rules, service instance in itself doesn’t do anything.

Encapsulation
Encapsulation dot1q defines the matching criteria of incoming and outgoing frames. We can match single vlan, double tagged vlans, class of service (CoS), ether types and layer 2 protocols under any EFP(Service Instance).

Bridge Domain
A bridge domain is single broadcast domain which could be used to mapped two different service instances into a single one. Bridge domain is identified by the service instance number. It has nothing to do with the vlan. So the service instances in same brige domain can communicate with each other.

Split Horizon
Service instances which are in same bridge domain can communicate with each other but by adding split-horizon they cannot forward the traffic from one instance to another. This feature is used when the same vlan is used for the instances but both belong to different customers.

Rewrite
Rewrite is used to modify the vlans packets by adding or removing dot1q tag.


Click Here To Read Rest Of The Post...

What is Service Instance In Metro Ethernet?

Legacy interface gives the flexibility to create sub-interfaces which is also known as logical inteface. Every logical interface consumes one IDB which has the limitation of scalability factor of vlans in the box. But the Metro Ethernet architecture which is EVC comes up with new thing which is service instance. It gives the many more functions with addition to the legacy logical interface. Service Instance could be considered as a service which can use a single port with the combination of layer 2 and layer 3 ports. The data entering in physical ports matches the service instance rules and filters the data according to the rules. We can create different type of service instances and bind all the service instances on the single physical port which gives the flexibility to service provider in terms of scalability. The another advantage of using service instance is that one can bridge the different domains but the same functionality could not be used by the sub-interfaces.

Different Type of Service Instances:-
1. Single Dot1q tag matching
2. Double tag matching
3. Untagged or native packets matching

Click Here To Read Rest Of The Post...