I’ve seen a lot of back-patting about Appgate’s security model, especially around the whole "zero trust" thing with TLS inspection. But let's be real—adding another layer of decryption/re-encryption isn't free. It's magic until you have to pay the mana cost.
In our PoC, we noticed a slight but perceptible lag on high-volume API calls between our microservices when the inspection was cranked up. Nothing catastrophic, but enough to make the performance purists twitch. The sales engineer mumbled something about "modern hardware" and "negligible impact," which is the kind of vague reassurance that sets off my spidey-sense.
So, before we commit, I'm digging for actual numbers. Has anyone done—or seen—benchmarks quantifying the overhead? I'm talking:
- Latency added per request (especially for east-west traffic)
- Throughput ceiling changes on your gateways
- CPU/memory hit on the controllers under load
- Any weird interactions with specific TLS versions or cipher suites
Anecdotes about "it felt slower" are welcome, but I'd kill for someone who actually measured before/after in a controlled environment. Bonus points if you compared it to a model without inspection or to a competitor's approach.
Or, you know, tell me I'm overthinking it and the overhead is truly a rounding error. I can take it.
But what about the edge case?