# Network architecture with primary/backup gateways in different locations **URL:** \ **Category:** Ignition\ **Created:** [May 17, 2023, 4:45pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849 "2023-05-17T16:45:05Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![Ryan\_Allen](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/ryan_allen/32/47013_2.png) [@Ryan\_Allen](https://forum.inductiveautomation.com/u/Ryan_Allen)\ **Post date:** [May 17, 2023, 4:45pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/1 "2023-05-17T16:45:05Z")
Does Ignition have a viable network architecture setup similar to this one from VTScada? ![image](https://us1.discourse-cdn.com/inductiveautomation/original/3X/2/e/2efb23ba653923bd037f4fd370a6b89ebef6d3a2.jpeg) I'm thinking this one is the closest that I can find mentioned, can anyone confirm this would work for the application above? ![image](https://us1.discourse-cdn.com/inductiveautomation/original/3X/c/2/c219e708f8a5691525a7338423d16b1c5d69ffe4.png) ---
**Author:** ![pturmel](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/pturmel/32/3295_2.png) [@pturmel](https://forum.inductiveautomation.com/u/pturmel)\ **Post date:** [May 17, 2023, 4:51pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/2 "2023-05-17T16:51:58Z")
Such a system would be vulnerable to split-brain on comms loss. Any device polling over the VPN would likely be pathologically slow. I don't see how VTscada would avoid the same issues. ---
**Author:** ![Ryan\_Allen](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/ryan_allen/32/47013_2.png) [@Ryan\_Allen](https://forum.inductiveautomation.com/u/Ryan_Allen)\ **Post date:** [May 17, 2023, 4:57pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/3 "2023-05-17T16:57:54Z")
Do you know of a good way to accomplish something similar? What if the two plants were connected by fiber and the vpn wasn't necessary? Split brain is still too much of a problem? ---
**Author:** ![pturmel](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/pturmel/32/3295_2.png) [@pturmel](https://forum.inductiveautomation.com/u/pturmel)\ **Post date:** [May 17, 2023, 5:01pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/4 "2023-05-17T17:01:06Z")
Split-brain isn't a latency issue, but a conflicting master issue when comms are out longer than the switchover delay. When comms goes down, you aren't going to be able to poll the other site anyways, though. I'd just use two normal Ignition instances. I wouldn't physically separate an Ignition redundant pair unless there was a **redundant comms path** between the two sites, and that comm setup's "healing" delay was substantially less than (2x ish) the Ignition fail-over delay. ---
**Author:** ![Ryan\_Allen](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/ryan_allen/32/47013_2.png) [@Ryan\_Allen](https://forum.inductiveautomation.com/u/Ryan_Allen)\ **Post date:** [May 17, 2023, 5:14pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/5 "2023-05-17T17:14:58Z")
> [@pturmel](#): > > Split-brain isn't a latency issue, but a conflicting master issue when comms are out longer than the switchover delay. Apologies for the confusion, I was just wondering if we eliminated the slow polling downside to this architecture, would the split-brain issue still be enough to recommend against it? Sounds like the answer is yes. How would you recommend having both the WTP and WWTP info displayed at both plants? Gateway to gateway communication with no expectations of any redundancy? ---
**Author:** ![pturmel](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/pturmel/32/3295_2.png) [@pturmel](https://forum.inductiveautomation.com/u/pturmel)\ **Post date:** [May 17, 2023, 5:17pm UTC](https://forum.inductiveautomation.com/t/network-architecture-with-primary-backup-gateways-in-different-locations/73849/6 "2023-05-17T17:17:41Z")
> [@Ryan\_Allen](#): > > both the WTP and WWTP info displayed at both plants? If useful to the operators, sure.