# Ignition Edge/Gateway Architecture Questions **URL:** \ **Category:** Ignition\ **Tags:** architecture, perspective, edge, ignition83\ **Created:** [October 10, 2026, 5:35pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417 "2026-10-10T17:35:39Z")\ **Posts on this page:** 8\ **Page:** 1
**Author:** ![r123](https://avatars.discourse-cdn.com/v4/letter/r/838e76/32.png) [@r123](https://forum.inductiveautomation.com/u/r123)\ **Post date:** [October 10, 2026, 5:35pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/1 "2026-10-10T17:35:39Z")
The site has 10 different buildings, each with a dedicated PC. Initially, we were told there would be no interconnectivity between buildings. Because of licensing costs, the original thought was to use Ignition Edge Panel for each building. I believe it would only support two connections at once, which I think would be fine, but it would then be limited to 35 days of historical data collection and possibly have some other limitations. Ideally, if we had full connectivity between the buildings and all the devices in them, I would set up a single Ignition gateway, with all the PCs in each building running only a Perspective View client. They are now talking about setting up a separate network where only the PCs in each building would be able to communicate with each other. Each PC would have multiple network cards: one to communicate with all the devices in that building and another to communicate with the other PCs. I'm not sure if that would mean we would need something like Edge IIoT on the PCs in each building to gather I/O from the local devices and then share that data over the dedicated PC network with a single gateway. I'm also not sure whether the network would be set up in a way that would prevent a single gateway from accessing devices in each building that would otherwise be inaccessible. For redundancy, I was considering installing Ignition Edge Panel on each PC. However, the Perspective screens for each building would need to read their I/O locally while pointing to the central gateway for historical data. If we were pushing out a single application using EAM to manage it across all the PCs, I'm assuming we could configure the screens in each building to read their data locally, but if an operator navigated to screens for another building, those screens would retrieve their data from the central gateway. I haven't tried doing that before, though. I've never used Edge Panel with Perspective before. One of the other engineers mentioned just the other day that some graphic objects or features aren't supported in Edge. I thought, or at least hoped, that Edge 8.3 supported everything that a full Ignition gateway does. ---
**Author:** ![michael.flagler](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/michael.flagler/32/26272_2.png) [@michael.flagler](https://forum.inductiveautomation.com/u/michael.flagler)\ **Post date:** [October 10, 2026, 7:46pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/2 "2026-10-10T19:46:59Z")
For the local PCs in each building, you'd install Ignition Edge Panel (which has all the features of IIoT but with the allowance of 2 clients per system). Edge does not allow connecting to remote tag providers though, so you'd only see the local tags when connected to the local project. You could connect to the central gateway's client to be able to see all systems, but if that network connection fails, then you'd have to launch a separate local client to still have visibility and control from the local PC. If you were using Vision, it has the capability of automatically doing this with a local fallback client that if it loses connectivity to the central gateway that it will connect to the local gateway automatically, but this capability isn't in Perspective. As for Perspective components, Edge is the same as Perspective except if you're trying to load 3rd party modules. There's some that have been approved/whitelisted by Inductive Automation, but they'd all require you to buy a license for them to use them on Edge. Otherwise the limitations are mainly that you can use database access, you're limited to 2 clients, history is limited to 35 days or 10M tags if I recall, and you can only have a single project and tag provider. There's no limitations for the built-in graphics/symbols, no limitation on scripting (besides things like DB scripting since it's not allowed), and no limitation on tags or devices. ---
**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:** [October 10, 2026, 10:13pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/3 "2026-10-10T22:13:48Z")
I would use a single redundant pair of Standard edition to cover the whole site. If the client is installing new circuits across the site, suggest the pull extra so you can run rapid spanning tree (or your favorite branded protocol) dual links to all buildings from the central server space. Much better bang for the buck than scattered Edge PCs. And will save you gobs of engineering for fallback projects. If the separate buildings have their own subnets, just make sure everything is routable to the central gateways. (Or use trunks at your servers and bring every subnet to the gateways as VLANs.) ---
**Author:** ![Chris\_Bingham](https://avatars.discourse-cdn.com/v4/letter/c/e5b9ba/32.png) [@Chris\_Bingham](https://forum.inductiveautomation.com/u/Chris_Bingham)\ **Post date:** [October 11, 2026, 1:12am UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/4 "2026-10-11T01:12:07Z")
> [@r123](#): > > I thought, or at least hoped, that Edge 8.3 supported everything that a full Ignition gateway does. In the context of core Perspective components, this is true. When Edge is not a consideration, many add (free) 3rd party modules to extend functionality beyond the core components…which is not available (for free) on Edge. Edit: Sorry, I skipped over @michael.flagler’s message stating the same thing & more. ---
**Author:** ![r123](https://avatars.discourse-cdn.com/v4/letter/r/838e76/32.png) [@r123](https://forum.inductiveautomation.com/u/r123)\ **Post date:** [October 12, 2026, 3:08pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/5 "2026-10-12T15:08:58Z")
Thanks for the suggestion. I want to clarify the network architecture because I'm not sure a centralized Gateway would have direct access to the equipment in each data hall. Each data hall has its own isolated equipment VLAN containing PLCs and other devices. The customer is considering providing an additional shared VLAN connecting the PCs in each data hall, but that VLAN would only allow communication between the PCs, not direct access to the equipment VLANs. Each PC would therefore have two network interfaces: one connected to its local equipment VLAN and another connected to the shared VLAN. My understanding is that a centralized Ignition Gateway would not be able to communicate directly with the equipment in each data hall without additional routing, firewall rules, or VLAN trunking, which the customer may not allow. Given those restrictions, would you still recommend a centralized Gateway, or would using Edge IIoT or a Kepware OPC Server on each data hall PC to expose its local tags to a central Gateway through the shared VLAN be the better approach? ---
**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:** [October 12, 2026, 3:34pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/6 "2026-10-12T15:34:28Z")
> [@r123](#): > > Each PC would therefore have two network interfaces This is **at least** as much of a security hole as properly firewalled routing. Arguably **worse** , as firewalls provide dramatically better access to indications of compromise. And if the PCs are Windows, I would argue that it is an outrageously acute hazard to your security. (See my rants on engineering malpractice.) ---
**Author:** ![Nol](https://avatars.discourse-cdn.com/v4/letter/n/8e8cbc/32.png) [@Nol](https://forum.inductiveautomation.com/u/Nol)\ **Post date:** [October 13, 2026, 5:51pm UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/7 "2026-10-13T17:51:44Z")
Is there a reason each data hall needs to be its own isolated network? Either way I'd suggest you find better network design, using dual NICs to route to two networks screams someone doesn't know what they're doing. Not trying to be rude. ---
**Author:** ![Chris\_Bingham](https://avatars.discourse-cdn.com/v4/letter/c/e5b9ba/32.png) [@Chris\_Bingham](https://forum.inductiveautomation.com/u/Chris_Bingham)\ **Post date:** [October 14, 2026, 4:42am UTC](https://forum.inductiveautomation.com/t/ignition-edge-gateway-architecture-questions/117417/8 "2026-10-14T04:42:45Z")
> [@r123](#): > > Each data hall has its own isolated equipment VLAN containing PLCs and other devices. If high availability is a concern, **my preference** is to ensure that there is no firewall between an Ignition gateway and the automation devices which are polled by that gateway (or redundant pair). If security is a concern, my preference is to enforce proper network segmentation (physical and virtual). If data hall networks **need to remain isolated** , I would entertain an Edge Panel (or pair) in each hall (polling their local devices), each connecting (thru a firewall) to a single standard gateway (or pair) which exists on the wider (V)LAN. If multiple clients per hall are required, or access to all data halls is possible from any data hall, then all clients connect to the standard gateway by default. A single client (or two) per data hall can/should have a client failover project configured (using Perspective Workstation) to auto-connect to local Edge Panel project if the wider LAN is down. If hall network segmentation is not a requirement, then a redundant / dedicated automation network topology with single standard gateway (or pair) starts to make a lot of sense, provided there is strong consideration taken to secure that local network if that gateway is accessible via WAN.