Hey everyone, I've been digging into ZPA for a possible implementation to replace our legacy VPN. The architecture with the connectors in our data centers/AWS makes a lot of sense for our zero-trust goals.
I've read the official datasheets, but I'm having a hard time translating the "up to 2 Gbps" throughput claim for a connector into something practical for planning. My experience with other SaaS tools is that the "up to" number is often under ideal lab conditions.
So, for those of you running this in production:
* What's a realistic, sustained throughput you've seen on a single connector (say, the large instance type) before you hit bottlenecks?
* Does the throughput limit apply to the sum of all connections through it, or is there also a concurrent connections limit that hits first?
* Does the type of traffic (bulk file transfers vs. interactive SQL queries) significantly impact the effective throughput you can achieve?
We're trying to size our connector deployment and budget accurately. I don't want to over-provision and waste money, but I'm really cautious about under-provisioning and creating a performance bottleneck for our teams. Any real-world numbers or scaling experiences would be super helpful!