The rapid development of broadband services to traditional telecom industry and the IT industry has brought profound changes. Multi-service, multi-network integration has become an irreversible trend. Broadband MAN MAN with the main network as the entity will be 3G, NGN and other new value-added services bearing platform.
3G, NGN such real-time voice and video applications require Metro to provide quality of service similar to traditional telecommunications technology and 99.999% carrier-grade network reliability. Meanwhile, fierce competition will drive the carrier to provide customers with guaranteed quality of service similar to SLA service, network reliability is one of the first and most important indicator. By improving the reliability of the network, operators can offer differentiated services, in competition between carriers occupies a favorable position to further establish and consolidate the corporate brand image.
MAN router reliability is reflected in the following two aspects, one is the equipment level of reliability, the other is the reliability of the network layer. Network reliability is the reliability of metro routers to protect an important element, because the traditional router slow convergence protocol (IGP at the second level, BGP level in the minutes), can not meet the needs of carrying real-time business. Network reliability is also a metro routers are active areas of new technology.
The current reliability of the emerging network layer IP routing technology, have fast convergence, end to end backup LSP, MPLS Fast Reroute, smooth reboot.
Fast convergence of IP routing
IP dynamic routing is the basic reliability of the network layer security mechanism is the inherent function of IP routing network. IP dynamic routing protocols for the network layer IP forwarding path calculation, the link or node failures led to the original data forwarding path is interrupted, the data forwarding path routing protocol for dynamic re-calculation, although a variety of routing protocols by using different mechanisms, The response time makes a difference, but the average level in seconds. Traditional IP services such recovery time is acceptable, but for real-time services such as multi-service bearer, carrier-grade IP network is required to restore millisecond response time, the traditional IP dynamic routing and a far cry from this requirement.
In traditional routing protocols based on IP routing to improve the agreement to shorten response time of failure, these measures are mainly to speed up the convergence of routing protocols. Speed up the convergence rate routing protocols can be divided into link failure detection, routing re-calculation, routing information updates, and several other considerations. By accelerating the link Hello messages sent between the frequency, speed up the SPF calculation speed and the route update message set high priority, routing protocols can quickly find the fault, and routed quickly and accurately update and speed up the convergence of routing protocols, By optimizing the IGP routing protocol convergence can be achieved is less than 1s.
Another way to speed up the convergence of routing protocols IGP and EGP is used to conduct a reasonable level of network planning, IGP for routing within the equipment, EGP (BGP4) carrying external routing, effective separation between the two routes, not each other re- distribution. IGP and BGP in the rational division of labor, forming a hierarchical routing structure, domain and inter-domain routing protocol convergence independent of each other, independently of each other, can achieve the fastest speed of convergence.
LSP protection switching
Protection switching is a term used by ITU-T, MPLS protection switching technology for improving network availability and stability is critical. Protection switching generally protected LSP Routing in the pre-calculation and allocation of resources, so you can guarantee that after LSP connection failure or interruption of network resources can be quickly regained.
The current technology can only support point to point LSP protection switching, protection can be used in two ways: 1 +1 protection and 1:1 protection.
Using a dedicated 1 +1 protection as the primary backup LSP LSP protection, in IngressLSR Department, the primary LSP and backup LSP bridge with the main LSP's traffic to a backup LSP at the same time be sent to EgressLSR, EgressLSR parameter value under fault indication , choose to receive traffic on the standby LSP.
1:1 protection is also a dedicated backup LSP protection as the main LSP, but the standby LSP is not the same for simultaneous transmission of traffic, the backup LSP LSP is working in the main premise can send other traffic flow protection switching decision in IngressLSR conduct.
MPLS Fast Reroute (FRR)
To meet such as video conferencing, real-time TV this class of business applications, must be provided for these flows is similar to traditional SDHAPS millisecond LSP protection.
LSP protection switching, signaling protocol required the intervention of the point of failure to restore the signaling transfer point of the fault indication introduces unnecessary network recovery delay. MPLS Fast Reroute technology can achieve in the absence of signaling involved in cases, the fault detection point directly to the failure link traffic based on pre-set path to redirect the protection, recovery points is the point of failure. Most of the Fast Reroute backup programs rely on pre-established channel, when the network detects a network failure recovery point, it is to be done is to simply update the LSP switching table, so that traffic from the port of LSP failure to switch to pre-establish the normal port within the LSP.
In addition to the advantages of fast re-routing can improve the protection and restoration of speed, through the weak links in the network options configuration protection, to avoid duplication in the reliable network protection, unnecessary consumption of the core network resources. MPLS Fast Reroute technology to provide protection switching within 50ms, can be used as an alternative protection mechanism SDHAPS.
MPLS Fast Reroute using the following configuration:
First, the entrance to the LSP LSR1, using a user command to activate MPLS protection switching function; LSR1 LSP path to send all the LSR signaling, are calculated for each LSR next hop LSR bypass a backup LSP, LSP Fast Reroute configuration is complete. When the LSP path to a downstream LSR detects a failure, by the LSR in the local traffic to switch to the backup LSP inside.
In the IETF in a variety of fast re-routing options, the two main ways to protect the link protection and node protection, the solutions to those problems and complexity of different, now the technology has not formed a formal RFC.
Smooth restart (Gracefulrestart)
Caused by control plane restart may include: software upgrade, software Bug or hardware failure, reset can be done without interruption control plane restart, the data forwarding plane without interruption. However, if the control plane failure, on the other routers will recalculate the route, bypassing the router failure, the data forwarding plane does not make sense without interruption, and failure route will spread to the entire network range. In MPLSVPNPE router If this happens, the result is disastrous.
Control plane technology can smooth restart to solve the problem, the router uses the technology in the control plane fails, you can inform the neighboring routers to continue to use the original path data forwarding, and restart the router and neighboring routers to re-establish routing state to ensure that the process of restarting in the business availability, a single device to minimize the impact of restarting the entire network.
In the smooth process of restarting the router does not save the agreement related to the state, so due to software failure does not continue to restart after restart.
Smooth resumption of the new features, many of the old equipment can not support, so it can support in the local subnet to use the features of the device.
The network border, the border router operators face many customers, and generally are not redundant measures, the most appropriate technology to use a smooth restart. Network core is generally protected using redundant paths, and restart with business likely to cause routing loop, so I do not recommend using a smooth restart the network core technology.