Hi everyone,
I’ve been using Teltonika gear for about two years now, mostly for basic remote site connectivity and some simple Modbus monitoring. I’ve always appreciated the “set it and forget it” reliability of the RutOS platform—it’s definitely saved me a few three-hour drives to reset frozen consumer-grade routers in the middle of nowhere.
However, I’m currently working on a new deployment that is a bit more demanding than my previous projects, and I’m hitting a bit of a performance wall regarding data throughput and VPN overhead. I’m hoping someone here might have experience with high-load scenarios.
The setup is an edge computing node that’s basically a high-spec virtualization host sitting in a ruggedized cabinet. The “brain” of the operation is an AMD EPYC 16-core processor running at 2.2GHz. We chose the 16-core EPYC specifically because we’re running several virtual machines for local AI inference and heavy logging. Because of the nature of the data, we need a permanent, rock-solid WireGuard or IPsec tunnel back to the main office for real-time synchronization.
The specific point I’m struggling with is the interrupt handling on the router side. When the 2.2GHz EPYC server starts pushing high-concurrency threads—especially during database syncs or when the VM backups kick in—the Teltonika gateway (I’m currently testing with an RUTX11) seems to struggle with the sheer number of packets per second. The CPU usage on the RUTX11 spikes, and we start seeing some jitter on the tunnel.
My personal insight from previous builds is that while we often focus on raw bandwidth (Mb/s), the real killer in industrial IoT is often the packet-per-second (PPS) count and how the gateway handles the encryption/decryption of that traffic when it’s coming from a multi-threaded server environment. I once tried to solve a similar issue with a cheaper gateway and ended up with a bricked unit after a thermal runaway; I’m determined to do it right with Teltonika this time.
I’ve already tried offloading some of the logging to a local MQTT broker to reduce the immediate strain on the tunnel, but I’m curious if there are specific “best practices” for tuning the firewall or the MTU settings when the router is paired with a backend as powerful as a 16-core EPYC. I don’t want the gateway to be the bottleneck for a server that has this much processing headroom.
Is it worth looking into the RUTX50 for the better CPU, or is there a way to optimize the current X11 to handle high-concurrency traffic from a professional-grade server more efficiently?
Also, has anyone noticed if certain encryption ciphers play nicer with the Teltonika hardware when dealing with sustained high-thread output from the server? I’d love to hear how you guys balance security overhead with raw throughput in these “pro-sumer” or light industrial setups.