Recently, the industry has proposed a new router algorithm, which is said to be very important in different types of large enterprise networks. This algorithm can automatically limit the number of network routes received by the router or update the link status to improve the efficiency of the router.
Previously, in the enterprise network, the oldest and slowest routers should wait for all other routers when receiving updates and recalculating their path tables. Professor Stephen Savage from the University of California in San Diego, who developed this algorithm with three other computer scientists, said that the new algorithm called XL (almost connected state) suppresses this update. Therefore, only the affected vrouters receive these updates. It is reported that he will introduce this paper at the computing machine Association's meeting on special interest groups in data communication.
Without the XL algorithm, a vro generally sends a large amount of route update information to the network, and each vro receives each update. In a super-large network, the number of routers and inevitable connection status updates often cause the routers to pause gradually.
The Savage says the update may only be related to the local region. He uses a map to illustrate the problem. He said a driver on the East Coast is not interested in whether the Highway 5 in Portland, Oregon is destroyed by floods. However, we tell everyone this information on the Internet.
To solve this problem, a large network needs to manually create some areas to isolate a group of routers in concept and limit the number of routers that can be accessed by flood information. The router is still receiving flood information, but only the router in the flooded area receives this information.
Savage said that the XL algorithm does not need to be manually set. Each vro automatically sets how other routers need to send this update so that this information can reach all destinations and prevent loops that generate a packet black hole.
The XL algorithm selectively suspends some updates and creates a trade-off. If a new link can be used after a fault occurs, this algorithm determines whether the information is sent to the nearest neighbor of the router to improve the path and ensure that the information passes through a sufficient proportion.
If not, the router will suppress the update and will not forward the update information. The result is that the update information is only sent to the area near where the structure has changed and the information publishing is rarely interrupted.
This benefit is balanced by the fact that using this algorithm means that each router lacks accurate information about the actual network conditions.
The router using the XL algorithm maintains the data of its neighbor's shortest path tree. How does its neighbor view the network and use the data to determine whether to update the forwarding path information. This will increase the total amount of data retained by the vro. However, Savage says his team thinks there is very little data to be added.
Savage says the slowest speed of routers in a large network limits the performance of the entire network. That is the router you are waiting for, so that the new network settings can cover every device.
Because the purchase cycle of a router in a super large network may be different, the old router with slow speed can have a great impact. If you purchased it 10 years ago and do not have the money to change the device, the performance of your network may be limited.
This algorithm is compatible with the protocol from the intermediate system to the intermediate system and the routing of the Open Shortest Path with priority. This means that software containing this algorithm can be deployed gradually. This software upgrade will be compatible with existing routing protocols. In these networks, the goal is to specify the Parameter Optimization Path, such as latency or bandwidth.
Savage said that the actual application of this algorithm requires the router vendor to integrate this algorithm into their own software. This requires vendor support. If Cisco applies this algorithm, it will have an impact. He has reported the algorithm to Cisco. Cisco has provided funding to help the study through the network system center.
Related Articles]
- Router principles, routing protocols, and Algorithms