SDN Enhanced VRRP based L2GW HA
Publication Date: 2015-Jun-24
The IP.com Prior Art Database
SDN(Software Define Network) L2(Layer 2) gateway was born to connect the SND overlay networking and legacy physical Lay2 VLAN devices, as SDN has been adopt by more and more customers, and these customers also have IT devices under traditional physical network infrastructure, there is no doubt that it’s a bottle neck of the communication between traditional network and SDN network. This invention introduces an Enhanced VRRP based L2 gateway (EVL2GW) HA(High Availability) that provides a traffic drop less failover mechanism to build High availability L2 gateway between traditional L2 and SDN network.
Page 01 of 6
SDN Enhanced VRRP based L2GW HA
Currently, SDN(Soft Define Network) has been adopt by more and more customers, and these customers also have IT devices under traditional physical network infrastructure, so to make sure that IT devices can communicatedwith server/vms(Virtual Machine) under SDN overlay networking environment, SDN L2(Lay 2) gateway was born to import traditional devices into SDN overlay networking in Lay2 VLAN method (not in IP layer, IP layer Gateway is out of this paper's scope.) The L2 gateway can be deployed as a virtual appliance or a hardware device, no matter it's a software of hardware, there is no doubt that it's a bottle neck of the communication between traditional network and SDN network, and it's one point of failure, it's a high risky node. To enhancethe HA(High Availability) of L2 gateway, VRRP has been involved. VRRP (Virtual Router Redundancy Protocol) is designed to eliminate the single point of failure inherent in the static default routed environment, but it(VRRP) only works on Layer3(IP-layer), not on Layer2(MAC-layer), so when involving traditional VRRP to L2 gateway, some incompatible issues willbe involved too, such as much fail over time in some case, see below picture, during L2 getaway fail-over, the packets from traditional networking to SDNoverlay networking will be dropped for a while(30secs in worst case).
Page 02 of 6
Figure 1 The problem of VRRP based L2 Gateway HA
Take above topology for example, the legacy serversin the legacy network communication with the VMs inthe SDN overlay network, two L2 gateways are working as active-standby mode for overlay networking side (only run VRRP on interface P1 and P2), assume L2 gateway 1 is master, and the active path from legacy server to VM1is "p13->p11->p7->p3->p1->vm1", if the master gateway 1 isdown, backup L2 gateway 2 will be the new master, but the network nodes in the legacy network are not aware the L2 gateway1 is down if the VLAN gateway1 is a virtualappliance, for example, in the Figure 1, the physical switch 1 can't detect any port link failure onP7, so the FDB(Forwarding Database) table still records the vm1's MAC address on p7, so packets from legacy server to VM1 will be forwarded to getaway1 via p7, path isalso "p13-> p11->p7->p3-> x", but at this time gateway1 is already down, then all these packets will be dropped for, at worse, 30 seconds.
Page 03 of 6
This invention introduces an Enhanced VRRP based L2gateway (EVL2GW ) HA that provides a traffic drop less failover mechanism to build High availability L2 gateway between traditional L2 and SDN network. When the L2GW master is down, the VL2GW will use the information from the GST to flush the legacy side network, and when the legacy side FDB is updated, the traffic will beshifted to the right path to the new master, and the traffic will not dropped (Refer to Figure 2). In order to achieve the drop less, the EVL2GW uses the GST, GSTMS and ENS to guarantee the traffic shifts...