gbproxy: Redundancy between NS-VCs (SGSN Side)
Each Server running osmo-gbproxy features multiple Ethernet Interfaces which can be connected to redundant Switches of a highly-available IP interconnect to the SGSN. The normal Gb/IP interface procedures of the IP-SNS ensure redundant NS-VCs are established via all of the Gb-Proxy side IP addresses.
We need to ensure that we continue to operate without interruption even when some of the NS-VCs to the SGSN are becoming unavailable.
#2 Updated by laforge about 1 hour ago
- % Done changed from 0 to 50
IP-SNS allows us to establish NS-VCs to multiple IP endpoints of each SGSN. In theory there is redundancy between those NS-VCs. We just need to test that this actually works.
Also, there are of course link-layer redundancy mechanisms on L2 or L3 such as STP, VRRP, OSPF, RIP, etc. that can be used to facilitate redundancy on the transport level between gbproxy and SGSN.
We'll lleave this ticket to extend our automatic tests to cover situations where some NSVCs are dying and traffic shall continue over other NS-VCs.