With every new firewall that we're purchasing, we're deploying the SandBlast Agent. At the moment we're only running it on about 20 firewalls, just because the licensing isn't retroactive. What we need to do is produce a proof of concept to say, "This is the stuff we're getting." We're looking at it in a learning mode and then we can consider getting into a more aggressive mode of stopping everything. At the moment, we're trying to use it to give us information rather than to fully stop everything.
It's deployed on our physical firewalls, on-prem.