Currently the backend will load graphs into memory unlimited, until out of memory occurs and the backend crashes.

The backend should have an option to set the max memory it is allowed to use for graph allocation. Currently each graph is cached as it stored, so that it doesn't need to reload a given graph if the frontend requests analysis for that graph. The load time for a graph is several seconds so its not desirable to always reload them.

The storage of the graphs into the backends application memory should use a least recently used algorithm for removing graph data to make room for new graph data in order to stay under the memory limit. However the algorithm to use is up for debate if any one has better ideas.