Detailed Explanation of Carrier Ethernet Testing Technology

1 Concept of Carrier

Ethernet As a fast, simple and high-bandwidth local area network, Ethernet has been used in enterprise networks for more than 30 years, but its application has been underestimated by telecom operators for many years. With the emergence of Carrier Ethernet (CE), operators gradually realized that in metropolitan area networks, Ethernet can be used as a data bearer network access, bear data services, and directly provide end-to-end dedicated line services.

The Metro Ethernet Forum (MEF) is the earliest industry alliance to study CE. The concept of CE was also proposed by MEF. According to the definition of MEF, CE has the following five major characteristics:

Scalability: Services are scalable, with the ability to allow millions of people to use the same network service; bandwidth is scalable, from 1Mb/s to 10Gb/s or even higher, and can be gradually increased according to a certain granularity. Carrier

-grade reliability: It can provide 50ms protection switching; it has end-to-end path protection capabilities; it supports aggregation link protection and node protection.

Quality of service: It provides guaranteed end-to-end performance; it supports the selection of end-to-end quality of service (QoS) levels; it is suitable for different applications such as business, mobile, and access aggregation.

Standardized services: can provide Ethernet private line and virtual LAN services; seamlessly integrate time division multiplexing (TDM) services; support circuit emulation services; support existing voice applications.

Service management: support fast service provision; support carrier-grade operation and maintenance management (OAM); can provide customer network management functions.

2 Testing of Carrier Ethernet

Currently, CE has become one of the hot topics in the industry. Many standard organizations such as ITU-T, IEEE, IETF, and MEF have started relevant standardization work, and there are many solutions on the market. Among them, the more representative technologies mainly include carrier backbone bridge/carrier backbone transmission (PBB/PBT) technology[1], transport multi-protocol label switching (T-MPLS) technology[2], enhanced Ethernet ring technology, and virtual private LAN (VPLS) technology[3]. So, how to judge whether these technologies meet the needs of CE? This requires starting from the five characteristics of CE defined by MEF, namely scalability, reliability, quality of service, service support capabilities, and service management, and conducting corresponding tests to evaluate whether these technologies meet these basic characteristics of CE.

2.1 Scalability Test

The purpose of scalability test is to verify the bandwidth and tunnel processing capability that CE can provide.

For bandwidth test, the test method of RFC2544 [4] can be used to test whether the tested solution supports line-speed forwarding of Fast Ethernet (FE), Gigabit Ethernet (GE) and 10 Gigabit Ethernet (10GE) interfaces.

The tunnel processing capability test mainly verifies the number of inner and outer tunnels that can be configured in the tested solution. When testing the number of tunnels, the tested solution needs to enable the tunnel protection mechanism. Only the test results that meet the 50 ms protection switching time are meaningful. The definition of tunnels varies for different solutions. For PBT technology, the outer tunnel is identified by the backbone network VLAN (B-VLAN) and the inner tunnel is identified by the backbone network service instance identifier (I-SID); for T-MPLS technology and VPLS technology, the inner and outer tunnels can be identified by the inner and outer multi-protocol label switching (MPLS) labels; for enhanced Ethernet ring technology, the inner and outer tunnels are identified by the inner and outer VLAN labels of the carrier bridging (PB) technology. During the test, it is necessary to verify the maximum number of tunnels supported by the tested solution and the maximum number of services supported under a single tunnel (i.e., the ability of multiple services to multiplex tunnels).

2.2 Reliability test

The purpose of the reliability test is to verify whether the tested solution supports a 50 ms carrier-grade tunnel protection switching time.

The protection switching time is the ratio of the number of packet losses to the packet transmission rate. There are many triggering conditions for the switching of end-to-end tunnels, including link failure, port failure, node failure, etc. These triggering conditions should be tested separately during the test. The corresponding operations may include unplugging the optical fiber, shutting down the forwarding port, shutting down the device power supply, unplugging the main control board of the device, etc. In order to simulate the real network conditions, when performing the protection switching test, a large number of tunnels should be configured according to the equipment conditions, a high data traffic should be loaded, and the traffic should be distributed as evenly as possible in different tunnels.

The network topology used in the reliability test can be ring, mesh or dual-home connection according to the technical implementation. When using ring networking, it is necessary to test the intersecting and tangent multi-ring topologies separately.

During the test, the protection switching of unicast tunnels and multicast tunnels should be tested separately to verify the protection of unicast services and multicast services by the tested solution. When

performing the protection switching test, it should also be verified whether there is packet loss when the tunnel is restored and switched back.

2.3 Quality of Service Test

The purpose of QoS test is to verify whether the tested solution can meet the service quality requirements of the telecommunications level.

The service quality test mainly includes access control policy, QoS flow marking policy, access control policy and queue scheduling mechanism.

The access control policy test is to test whether the tested solution can correctly configure the access control policy for the second, third and fourth layer traffic. At the same time, it is also necessary to verify whether the forwarding performance of the tested solution is affected after the ACL is configured. The

flow marking policy test is to test whether the tested solution can perform QoS marking or re-marking on the data flow according to the flow marking policy. QoS marking can be based on 802.3, 802.1ah, MPLS or IP service category (ToS)/differentiated services (DiffServ).

Access control policy testing is to test whether the device under test supports access rate control based on committed information rate (CIR) or peak information rate (PIR) for access traffic and the bandwidth granularity that can be controlled.

Queue scheduling mechanism testing is to test whether the scheme under test supports priority queue scheduling mechanisms such as priority queue (PQ), weighted fair queue (WFQ)/weighted round robin (WRR).

When performing quality of service testing, single-layer QoS policy and hierarchical QoS policy need to be tested separately. For quality of service testing, please refer to MEF14[5].

2.4 Service Support Capability Test

The purpose of the service support capability test is to verify whether the tested solution supports Ethernet dedicated line (E-Line), multipoint to multipoint Ethernet (E-LAN), tree Ethernet (E-TREE), time division multiplexing (TDM) and other service types.

When conducting E-LAN ​​service testing, the focus should be on the way the tested solution creates E-LAN ​​services and verify its learning method for remote media access control (MAC) addresses.

When conducting E-TREE service testing, it should be verified whether the leaf nodes can be isolated from each other, and the tested solution should verify the multicast function through E-TREE services.

When conducting TDM service testing, the delay, jitter and error performance of TDM need to be tested. The test port can select framed E1, unframed E1 and channelized STM-1 according to the device support.

In addition, the function of the tested solution supporting TDM clock synchronization should also be tested. The test content includes the clock synchronization function of the circuit emulation service (CES) service, the clock function of synchronous Ethernet and the IEEE 1588 time synchronization function. For testing of service support capabilities, please refer to MEF9[6], MEF18[7], and MEF19[8].

2.5 Service Management Testing

Service management testing mainly includes two aspects: testing of the network management system and testing of the OAM mechanism. The former should be based on the fault, configuration, billing, performance, and security (FCAPS) functional module, mainly including topology discovery, service creation, fault monitoring, performance management, etc. During testing, the services created by the network management system need to send simulated service traffic for verification.

The latter is based on standards such as IEEE 802.3ah[9], IEEE 802.1ag[10], ITU-T Y.1731[11], and G.8114[12]. Through a protocol analyzer, the message format and message flow of messages such as continuity check (CC), loopback (LB), and link tracking (LT) in the Ethernet OAM mechanism or T-MPLS OAM mechanism are checked for consistency to verify whether they comply with the relevant standards.

3 Practice of Carrier Ethernet Testing Technology

By using the above test methods, four CE implementation solutions, including PBB/PBT, T-MPLS, enhanced Ethernet ring network and VPLS, are tested. It can be seen that although the existing CE technologies can basically meet the needs of carrier-grade networking in terms of service carrying capacity, reliability, scalability, QoS and OAM management capabilities, it is difficult to achieve complete interconnection and interoperability in the short term due to the large differences between different CE solutions.

3.1 Service carrying capacity

From the test results, the above four solutions can all carry broadband Internet access services and IP voice services well, but for dedicated line services, multicast services and TDM services, different solutions have different implementation effects and complexities. When carrying dedicated line services, the enhanced Ethernet technology has a poor isolation effect on user information because all network devices need to learn the user's MAC address information, and the network is vulnerable to attacks from the user side; while PBB/PBT, T-MPLS and VPLS technologies only learn the user's MAC address information at the user network interface (UNI) of the operator's edge (PE) device, and the network core device does not learn the user's MAC address information, so the isolation effect is good. For multicast services, several solutions can carry them, but the implementation complexity varies. Among them, enhanced Ethernet technology can easily implement multicast services because it uses ring networking, while the multicast service implementation methods of PBB/PBT, T-MPLS and VPLS/EoMPLS technologies are more complicated, and there are differences in the specific implementation methods, and a unified standard has not yet been formed. For TDM services, currently PBB/PBT, T-MPLS and VPLS technologies can provide corresponding TDM interfaces or CES simulation interfaces, while the TDM support capability of enhanced Ethernet technology is relatively poor. For the synchronous clock transmission function of TDM services, most manufacturers are still in the process of research and development and cannot provide a relatively complete solution.

3.2 Service Reliability

For broadband Internet access services, IP voice services, and dedicated line services, several solutions can meet the protection switching time of less than 50 ms in the case of link and node failures; for multicast services, due to differences in implementation methods, some technical solutions cannot achieve a protection switching time of 50 ms, which is also one of the key points of the next technical research. In addition, the protection of access links is also very different. In addition to the standard Rapid Spanning Tree Protocol (RSTP) rapid spanning tree technology, many manufacturers have developed proprietary technologies to improve the efficiency of protection switching. From the test results, the damage time of upstream service traffic caused by RSTP fault switching is greater than 100 ms, while the test effect of proprietary technology is less than 50 ms, but because they are all private technologies of each manufacturer, their interoperability cannot be guaranteed.

3.3 Expansion Capability

At present, major CE equipment manufacturers have developed equipment that supports 10GE interfaces. The interface bandwidth can be expanded from FE interfaces to 10G interfaces, which can basically meet the needs of operators' metropolitan area networking. In terms of tunnel support capabilities, due to different equipment processing capabilities, the number of services it can provide is also different, ranging from thousands to tens of thousands, which can basically meet the needs of operators' metropolitan area Layer 2 networking. Among them, PBT, T-MPLS and VPLS have better scalability in this regard.

3.4QoS function

At present, several implementation schemes can basically realize flow classification based on VLAN tags and ports in terms of QoS flow classification capability and queue scheduling algorithm implementation, and support PQ, WFQ/WRR queue scheduling algorithms. For hierarchical QoS functions, most manufacturers can support it.

3.5Service management capability

At present, most manufacturers can provide a visual topology view and support service creation functions, but the configuration complexity is different. Most manufacturers' network management systems need to configure each network element separately, which is more complicated. Since the control layer of PBB/PBT and T-MPLS solutions does not support control signaling, service creation needs to be performed through static configuration on the network management system. For the configuration of batch services, most manufacturers can provide corresponding application programming interfaces (APIs) and implement them by writing scripts, but currently, various manufacturers basically do not support the creation of batch services through the network management interface, which also brings some inconvenience to the daily maintenance of operators.

In terms of OAM function support, most manufacturers now support the 802.1ag protocol, but only some manufacturers can implement the performance management function in the Y.1731 protocol.

4 Conclusion

Since different CE technologies are suitable for different application scenarios, when conducting specific tests, it is necessary to formulate corresponding test plans based on the application scenarios and the adopted technical solutions. With the development of various CE technologies, the test methods of CE technologies also need to be continuously improved to meet the growing evaluation needs of CE technologies.