The support team would be happy to discuss this and help get to the bottom of it. Your post doesn't share specifics on what kind of traffic you are seeing and the duration, but on-network devices should not be downloading through the CSA. If set to Dynamic these devices will check whether they can access the core and if there are any instances where the core is temporarily unavailable they will switch over to the CSA. Whatever may be the case, we're happy to dig into it with you and sort it out.
Understand / know your users.
There's "wonderful" situations where users keep disconnecting their laptops/tables from the corporate network & end up on the public WiFi (yes, I've seen this ... it happens) which can explain how you end up bouncing over the CSA.
In principle (as was pointed out) the "can I talk to home Core" is very simple. Things that can cause you grief here tends to be:
- DNS being crap / unreliable (the name can't be resolved ... try seeing if using an IP-based agent setting for the Core works better)
- Web / Network caching appliances (these cause all manner of unintended problems -- rarely in regards to the CSA though).
- "End-user caused issues" (see the above example) ...
- ... and generally anything that can / does touch your network & is part of infrastructure (Firewalls flaring up / load-balancers getting out of whack, etc. etc...)
It's very much not a general issue, and would need to be hunted down in a "your specific scenario" type of way. Everyeon's different (and so is everyone's network infrastructure with their own skeletons in the closet) ... so not a general thing that can be answered here like that.
But 4.000+ nodes in and of itself on a CSA is fine.
Thank you both for your reply. We have actually worked with Support since implementation and even paid for an additional 3 days of onsite to help figure out why its occurring. If it were all laptops I could see it but a lot of these devices are desktops which don't connect to wifi and shouldn't be dropping network connectivity throughout the day. We are talking about over 1000 hits an hour from on-net devices. We have heard from some engineers that we should expect to see some on-net traffic checking in to the CSA - and some have said we shouldn't be seeing the amount we are seeing - so our next step was to talk with other customers with similar setups to see what they are seeing in their environments.
If it's the same devices that are hitting the CSA so many times per day, it should be possible to setup some monitoring. As Paul pointed out, there are a lot of variables that can cause the device to think it doesn't have a connection to the CSA. It's tricky to tell you exactly what without expert knowledge of your network.
My immediate suggestion, specifically for your desktops, is to set your CSA connection differently. Instead of dynamically determining the path, you can set it to always connect directly. With this change, it will never attempt to connect via the CSA. If there are intermittent network issues, you may see some task failures, but this is a good thing because it will help you identify what is happening for these devices or the network.