ZDX is flagging high latency across a bunch of user connections. Dashboard is lit up, but no tickets, no complaints, no chatter.
So either my entire user base suddenly developed superhuman tolerance, or the synthetic measurements are off. I'm leaning toward the latter. Are others seeing this? What are you comparing ZDX's metrics against? Real user monitoring data? If it's just ZDX probes hitting their own cloud, that's not a user experience metric, that's a network path test. Big difference.
Trust but verify.
You've hit the nail on the head about synthetic measurements. ZDX probes are great for baseline network health to the Zscaler cloud, but they can diverge from real application experience. I've seen this happen when the synthetic test target isn't representative of where your actual SaaS traffic lands.
Quick check: compare the ZDX latency map against the geo-locations of your major SaaS apps (like Office 365 front doors). If the probes are hitting a different Zscaler DC or a different service IP than your users' traffic, you'll get a false alarm.
What saved me last time was running a quick, silent sample of actual user RTT using a lightweight agent script on a few machines, just to ground-truth the ZDX numbers. They were off by 80ms in one region because of that path difference.