Introduction to the virtual routing cluster technology of gigabit exchange Routers

I. Introduction:

Currently, IP has become the routing protocol for most backbone network products. In some network environments, users have high requirements on the network, and any shutdown and storage and transportation losses will have a serious impact on users. For example:

1. Internet service providers provide Web host devices. To make your Web server always valid for the public, 99.9999% of the normal running time of the user must be guaranteed.

2. The process control application must be able to access its control system in a timely manner; otherwise, serious control may occur;

3. Sometimes, applications running on the IP host will time out. If the service runs requires a high level of network applications, this timeout will have a bad impact.

More and more IP hosts use DHCP to specify their IP addresses. However, many IP hosts use manual configuration as the only means to discover Gateway Routing addresses. Some hosts use the gateway detection method, but this method is generally not recommended (RFC1122). Dynamic Ping to the gateway is also prohibited. The ICMP routing discovery protocol allows routers to be discovered through IP hosts, but it is not widely used.

This means that most hosts cannot quickly know whether the connection between the vro and the connected lan has failed, and it takes a long time for the IP host to detect the connection failure and exchange the replacement vro.

Therefore, it is important for any device to improve the network efficiency. Using vro cluster technology can effectively solve some problems. This article introduces the SSR Gigabit Exchange Routing cluster technology of CABLETRON for some discussion.

2. How to do it?

The vro cluster technology can be used to define multiple SSR routers in the same LAN as clusters for Route backup between each other, the SSR router in the cluster uses the IETF virtual redundancy Routing Protocol to implement Images Between route switches and the LAN of Each router. If a router or the LAN connected to it fails, other routers will automatically replace the failed router to bypass the fault point and re-route the route. The route recovery time is within several seconds. Therefore, this is a transparent application running on the IP host.

(Figure 1)

1. VRRP Concept

As shown in figure 1, VRRP can provide a backup router when the default gateway of the workstation fails. VRRP can create a vro with a virtual MAC address and virtual IP address.

It is true that we can add multiple default gateways in Windows 98, but this feature can only be used when the host is started. In other words, when the host is started, the first default gateway in the gateway list uses ARP for address resolution. If the resolution fails, Other gateways in the list continue to perform address resolution.

What happens when the host finds the MAC address of the default gateway? You can only restart the host to continue searching for the next default gateway.

VRRP can be used to solve this problem because it is dynamically changing. If the primary router is paralyzed, the backup router automatically replaces the primary router. The host is unaware of any difference.

The virtual routing cluster technology is integrated into SSR as a module to ensure the practicability of network applications and improve the reliability and adaptability of networks with high application requirements.

CABLETRON's virtual routing cluster technology complies with the VRRP (Virtual redundancy routing protocol) standard protocol developed by IETF (Internet Engineering Task Group.

　　
(Figure 2)

In Figure 2, IP host H1 and H2 are connected to a LAN with multiple routers (R1, R2, etc.). R1 and R2 are connected to the destination D1. Under normal circumstances, only the routing IP addresses of a single router, such as H1 and H2, are configured. The problem is that when any router fails, one or more hosts, such as H1 and H2, are connected to the destination D1. For general routing protocols such as RIP and OSPF, it may take a long time (about 40-90 seconds) for host H1 and H2 to discover a route failure, or even cause a TCP connection interruption. When VRRP is used in vro cluster technology, it takes only a few seconds to restore the failed route in the network. This greatly improves the network efficiency, provides route redundancy for network applications that require picky backbone.

2. How does the vro cluster technology work?

Each vro in a vro cluster is assigned a virtual MAC address corresponding to the corresponding IP address. Then, configure each SSR as a route with the MAC address and IP address of all other routers. The IP host is configured to use any or all vrouters as its own vrouters.

SSR selects the master route using VRRP information in the cluster, and other SSR becomes the hot backup router. Each vro in the cluster provides a virtual MAC address (instead of its actual MAC address) as the original address for ARP (Address Resolution Protocol) to respond to the host, each vro also sends IP host jobs optimized by ICMP (inter-network control packet protocol) redirection. If any router in the cluster or their LAN connection fails, the master router takes over the corresponding job and replaces its virtual MAC address and IP address to implement ARP response. If the primary router fails, the remaining router will select a new primary router, which replaces the virtual MAC address and virtual IP address of the failed primary router.

3. ICMP route discovery and ICMP redirection

ICMP routing allows IP hosts to use ICMP information and programs to differentiate routers. Use ICMP and SSR to periodically broadcast ICMP route broadcast information and respond to ICMP route requests from IP hosts.

ICMP redirection is the best Routing Mechanism for the router to notify the IP host to reach a specific destination. ICMP redirection follows the RFC1122 standard.

The virtual routing cluster uses the standard ICMP redirection mechanism to control the router to send IP packets to the destination by routing around the link failure point. The information package is still forwarded by SSR before redirection, so that data will not be lost due to route changes.

　　
(Figure 3)

Figure 3 shows how a vro cluster works. In this chart, both vroa A and vrob B are components of the same vro cluster, and each of them is set as the primary router in its own cluster, while other clusters are set as backup routers. Vroa A is configured as its own IP address and virtual MAC address (in the text area in figure 3), and the IP address and virtual MAC address of cognitive vrob B (in the Italic area in figure 3 ). Similarly, vrob B is configured as its own IP address and virtual MAC address (in the positive text section in figure 3), and the IP address and virtual MAC address of cognitive router A (in the oblique text area in figure 3 ).

In this case, ssr a is selected as the primary router, and ssr B is selected as the backup router accordingly. Both routers can send ARP requests to host 1, 2, and 3 through their virtual MAC addresses. Both routers can also route traffic from IP hosts 1, 2, and 3.

Iii. Results:

1. What will happen if an SSR router fails?

Assuming that ssr B is disconnected from the local area network or the serial port module circuit breaks down, as shown in 3, ssr a quickly discovers that ssr B fails and quickly obtains the IP address and virtual MAC address of ssr B, ssr a now takes the initiative to respond to ARP requests sent to failed vrossr ssr B, and obtains the virtual MAC address of the failed vro as the original address, and actively receive IP data sent to the virtual MAC address of the failed router, and forward it to the corresponding destination.

　　
(Figure 4)

2. What will happen after the SSR router is restored?

If ssr B recovers, it will notify itself and SSR to stop imitating it, And then ssr B will respond to ARP requests with its own MAC address, the two routers route information packets at the same time. There is only a short interval between the recovery response of ssr B and the virtual address that the primary router stops using. During this short period of time, data packets may be lost or copied. Generally, the TCP transmission protocol can ensure data recovery at the end.

Iv. Conclusion

Generally, it takes several minutes (or even no) to use RIP and OSPF in a system to detect a fault ). However, after configuring a route in SSR, you can use VRRP to configure a certain number of virtual Routing Clusters to detect and confirm a fault within five seconds, and quickly replace it with a new route. This recognition and re-routing capability greatly improves the fault recovery capability of backbone network applications. On a CISCO Route switch, hot route backup (HRVP) can be used for implementation. Many other products have implementation methods.

[1]

Article entry: csh responsible editor: csh