Group Mobile return is not a layer-2 Exchange

Meanings of L2 Switching

Ethernet is the most commonly used protocol in the home and office so far. The connection is very simple. You only need to connect the computer to the Home Office switch or the RJ45 socket on the office wall. To allow a data group to find its destination, multiple protocols and technologies are running on both the data plane and the control plane. These are collectively referred to as L2 switches or bridging. The basic features of data include:

1. Forward Forwarding Based on MAC Destination Address and VLAN ID

2. MAC source address learning and aging

3. Unknown MAC address flooding

4. multicast and broadcast

5. VLAN member set

6. VLAN processing (insert/replace/remove service tags and customer tags)

Port filtering or isolation

Ethernet Switching devices support most of the above functions. These devices are widely used in network devices for group switching. Network Device designers are familiar with switching devices and how to use them.

We are now inGroup Mobile returnIn the age of group exchange, carriers are building a group exchange network to transfer groups from base stations to RNC. The switch mode of the group is different from that in the LAN environment, and the standard Ethernet switch device does not provide sufficient functions. A common technology of group mobile return is the group transmission network (PTN ). This article introduces some significant differences between L2 switching and PTN switching.

Forward

The standard L2 Switch will forward one group from one Ethernet port to another. Switch between physical ports.

The PTN switch on the network edge has two aspects: user interface and network interface. The user traffic that the user interface connects to the Ethernet port. The network interface is connected to the network device through a pseudo-line (PW) and a tunnel (LSP. Many PW ports lead to the same LSP, while many LSPs may lead to the same physical port. Groups are exchanged between user ports or between user ports and PW. Although the forwarding method is similar to that of a L2 Switch, PTN adds a new dimension, namely, the virtual PW port.

Encapsulation

In layer-2 switching, the Group format is usually unchanged. Most group editing tasks are only performed in VLAN fields. The inbound and outbound groups can carry 0, 1, or 2 VLANs. VLAN processing rules must be flexible enough to process any combination.

For PTN switching, the user group must be encapsulated into the PW load after being switched to the PW port. PW groups can have very complex formats, including external L2 headers, external VLANs, LSP tags, PW tags, control words, and internal MAC or IP frames. The switch must first classify external L2 headers and tags to identify PW and PW-tagged protocols. Then, the Internal Frame (Ethernet or IP) can be extracted or forwarded. This MAC-MPLS-MAC format poses a severe challenge to grouping processing. Some features (such as classification) need to be used twice. For the first time, they are used for external MAC and then for internal MAC. Some pipeline-based architectures may not meet these requirements.

Forward table

The L2 Switch uses the MAC Destination Address and VLAN to search for the forward table. Generally, a L2 Switch has only one table.

The PTN switch requires multiple forward tables to support multiple users. The search will be divided into two steps. The first step is to identify the forward to table, while the second step is to locate the forward to table rules.

Protection

The nature of Ethernet is "Best Effort". Therefore, protection is not a concern. This is good enough for the LAN environment, but not for the carrier network. Carriers are used to the APS (Automatic Protection Switching) support capabilities in SONET/SDH and apply this idea to PTN.

First, PW and LSP require and 1 + 1 protection. When the primary PW fails, all traffic in the PW must be switched to the standby PW within 50 milliseconds. Similarly, when the primary LSP fails, all PW In the LSP will be switched to the standby LSP within 50 milliseconds. Then, ring network protection requires that all traffic on the Ring Network be re-routed when the ring network fails. For the mesh network technology, FRR is also a protection mechanism to ensure fast recovery in case of link failure.

All these protection mechanisms require the data plane to check the protection status and make a forward decision. Group replication is required under 1 + 1 protection. CPU intervention should be completely avoided, or the lowest possible level should be maintained. This poses a challenge to the design of data paths.

OAM

Like protection, OAM is another feature that traditional L2 exchange ignores. From the operator's perspective, this is a required feature. CFM and PM are required for both PW and LSP levels to support APS and ensure compliance with SLA ). Connection check (CC) is the key to detecting any faults and triggering APS. Latency and loss measurements are important evidence to meet SLA requirements.

In the past, OAM functions were separated from data plane processing and processed separately. The trend is that OAM is becoming more and more integrated with data. For example, CC groups should implement as many fault detection times as possible. However, a higher grouping rate requires higher performance support, so it is best to process CCM on the data plane. Latency measurement (DM) requires hardware to be able to insert timestamps. The loss measurement (LM) utilizes the data path statistical results. It is of great significance to integrate DM and LM into the data plane design, which puts forward new requirements for the data plane processor.

Layer 2 and Layer 3

In the past, vswitches were switched before the Layer 2, while vrouters were switched before the Layer 3. These are different devices and devices used in different networks.

PTN is a service type. The carrier certainly hopes to support Layer 2 and Layer 3 services in the same group network. Therefore, Ethernet and IP network traffic from user interfaces will be accepted and transmitted through PTN. For this reason, the PTN switch must be able to provide Layer 2 and Layer 3 services. For layer-2 services, it switches Ethernet groups to PW Based on the forward-to-database. For a layer-3 service, It routes IP groups to PW Based on the route table. The ability to process layer-2 switching and layer-3 routing is a huge difference.

Horizontal segmentation

As we all know, the PTN switch performs group switching between the user port and PW. Generally, this must follow some additional forwarding rules. For example, the forward forwarding between two PW networks is not allowed in a mesh network. In addition, forwarding from one user port group to another user port group is not allowed. This forward forwarding rule based on PW or port is the basis of horizontal segmentation.

Service quality

Service quality is a set of features used to enhance SLA compliance. It may include policies, integer types, schedules, traffic management, congestion control, and comments. In short, compared with L2 switches, PTN switches require more comprehensive and advanced service quality features.

Synchronization

In addition to data, the carrier also wants to transmit time information to the base station. The time information may be the frequency, stage, or time of the day. The IEEE 1588 protocol is a good alternative protocol to achieve this goal. However, the success of IEEE 1588 depends largely on the group network, especially the network latency changes. If the 1588 group encounters excessive latency changes or asymmetric latency, it cannot restore accurate time information from the 1588 group. There are two ways to solve this problem:

(1) When designing the network, it is said that the Latency Change remains extremely small.

(2) supports transparent clock on each network node. In this way, the network latency is known and can be compensated.

In many cases, the 2nd options are easier to implement, which lowers the requirements for PTN to support transparent clock. From the hardware point of view, the switch needs to calculate the wait time of the 1588 group in the switch, read a value from a dedicated field in the group header, and add the wait time to the value, and write it back to the group. All of these are performed during group transmission.

Conclusion

We have discussed L2 switching and PTN switching, and identified many key differences in features and functions. There are many differences between the two. L2 switches are mature and well known by network equipment designers. PTN is a new technology for grouping mobile return, with bright prospects. Although both are based on group exchange, the difference is also obvious. A ptn can be considered as a superset. It basically includes all the functions of a L2 Switch, but also adds more functions.

In terms of group mobile return, operators need not only layer-2 exchanges. They also require more powerful and flexible data surface processors to meet these requirements. Since PTN and other technologies are still new technologies and many protocols and standards have not yet been fully recognized, it is important to choose a programmable and Scalable Data plane solution.