Method and device for solving unified network element scheduling of large-scale network

Abstract

The invention discloses a method and device for solving unified network element scheduling of a large-scale network, and the method comprises the steps: a unified scheduling service transmits an acquisition control request of the same equipment to an NED instance group, and the NED instance group is obtained through dynamic calculation of an enhanced consistent Hash algorithm; and selecting an optimal NED instance from the NED instance group through a load feedback mechanism to process the request. The method and the device are based on an enhanced consistent Hash algorithm and a load real-time feedback mechanism, on one hand, the request success rate is improved, on the other hand, scheduling is reasonable, and the load balancing problem is solved.

Description

technical field [0001] The invention relates to the field of communication networks, in particular to a method and device for solving unified network element scheduling in large-scale networks. Background technique [0002] The number of communication network devices is huge, with thousands of devices at the backbone layer and hundreds of thousands of devices at the access layer. The network management system usually has a device driver layer when connecting devices for interaction. Other application services of the system, in order to shield the differences between manufacturers and access methods, and solve the problem of large-scale procurement and control. [0003] The usual way to solve large-scale procurement and control is to use the device driver service for horizontal expansion to balance the load. The ratio of the number of devices to the driver service instance is about 500:1, but this method is not completely applicable to datacom devices. There are mainly the f...

AI Technical Summary

Problems solved by technology

[0005] 2. The number of vty (virtual terminal) of the device is generally limited. For example, the maximum vty of a user is 5, that is, once the 5 connections are full, the new request will fail because the device connection cannot be obtained.

[0006] 3. It is expensive to frequently establish a connection with the device. Usually, it is necessary to put the device connection in the connection pool at the driver layer for ready use at any time.

[0007] like figure 1 As shown, when the upper layer application (AppCol) interacts with the device, according to load balancing, it will randomly select a device driver service instance (NED instance), and the device driver service instance (NED instance) will try to establish a connection (pool) with the device, and The connection (pool) is temporarily saved in its own connection pool; but when the request increases, all available connections of the device may have been occupied by the device driver service instance (NED instance), and when there is a new device driver service instance (NED instance) to obtain it fails when connecting

Images