You will notice that physical memory usage tends to be almost always at full capacity, with a very quick increase to that level when you reboot your server.
This is normal and is why Server Density also reports cached memory. Very simply, Linux does not release the memory as far as the monitoring is concerned because the data is cached ready to be accessed much faster the next time it is needed. Linux manages the memory so that if a process needs more, then it will be allocated appropriately, clearing the cache where necessary.
What you need to watch is swap memory. Swap uses space on the hard disk to act as virtual memory. When Linux runs out of physical memory, it will start "swapping". This is generally bad because hard disks are much slower than physical RAM and you will start to see performance decreases as swapping increases. If you are seeing a large amount of swap space being used then you should drill down to the process level in the server snapshot by clicking on the graph data points to find out exactly which processes were eating up the memory. It may be necessary to tweak the service config or increase the available RAM.
Note if you are using AWS, that most EC2 instances don't have swap memory, so this won't be a useful alert for you ;)
This plugin adds a more useful "AvailabeMemory" metric: https://plugins.serverdensity.com/freememory/