High-availability cluster architecture and protocol

Abstract

Methods and systems are provided for an improved cluster-based network architecture. According to one embodiment, an active connection is established between a first interface of a network device and an enabled interface of a first cluster unit of a high availability (HA) cluster. The HA cluster is configured to provide connectivity between network devices of an internal and external network. A backup connection is established between a second interface of the network device and a disabled interface of a second cluster unit. While the first cluster unit is operational and has connectivity, it receives and processes all traffic originated by the network device that is destined for the external network. Upon determining the first cluster unit has failed or has lost connectivity, then all subsequent traffic originated by the network device that is destined for the external network is directed to the second cluster unit.

Description

COPYRIGHT NOTICE[0001]Contained herein is material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction of the patent disclosure by any person as it appears in the Patent and Trademark Office patent files or records, but otherwise reserves all rights to the copyright whatsoever. Copyright ©2013, Fortinet, Inc.BACKGROUND[0002]1. Field[0003]Embodiments of the present invention generally relate to the field of computer networks. In particular, various embodiments relate to methods and systems for an improved cluster-based network architecture that enables efficient load balancing and failover protection between units of a cluster.[0004]2. Description of the Related Art[0005]In the field of distributed computing, two or more computing resources are generally used to perform various tasks such as traffic management, synchronization, load balancing, and failover, among other such tasks. These grouped resources generally form a computi...

AI Technical Summary

Benefits of technology

[0015]Methods and systems are described for an improved architecture and design of high-availability clusters enabling efficient load balancing and failover protection between units of a cluster. Methods and systems are also described for connecting network devices with cluster units such that load from network devices is balanced across cluster units and failover provisions allow efficient allocation and sharing of cluster unit resources. Network devices can include routers, switches, or access points that allow routing or flow management of traffic and can even include client computing or web-enabled devices such as personal computer or laptops that are directly / indirectly connected with cluster units. Cluster units, on the other hand, can include routers, gateways, servers, among other such devices that support synchronization of content across cluster and ensure continuity of traffic flow to and from the Internet.

Problems solved by technology

Existing cluster architectures require a master-slave configuration, which, apart from creating other inefficiencies, demands all network devices to be connected to the master cluster unit, thereby increasing the load on the master unit and not allowing balancing of load on all cluster units.

Further, in an active-passive HA cluster mode, as computing devices in a network are connected to a single master cluster unit that manages data traffic, when the master cluster unit fails, there exists no other master cluster unit, and a delay is created before the slave cluster unit is assigned the role of a master cluster unit, leading to a potential loss of data or other relevant configuration information.

Active-active HA cluster mode method also faces similar problems as of active-passive HA cluster mode method as the process of selecting and assigning new primary cluster unit takes time and creates load balancing issues in the cluster.

Furthermore, existing architectures require slave cluster units to re-learn IP / MAC after every failover, which again creates inefficiencies and delay.

Images