Same as anything else -- you start by adding a few more front-end servers (or VMs), or scaling up the ones you have just a bit more. You don't have to do nearly as much work, and not nearly as soon.

If load levels continue to grow, leverage ehcache to handle caching directly in the VM (if you're not already), begin splitting responsibilities across different service implementations, etc.