More Issues for VM Paging
1. synchronizing shared pages
2. clustered reads/writes from backing store, and prefetching
3. adapting replacement strategies (e.g., switch to MRU)
4. trading off memory between the file (UBC) and VM caches
5. trading off memory usage among processes
6. parameterizing the paging daemon:
Keep the paging devices fully utilized if pages are to be pushed, but don’t swamp the paging device.
Balance LRU accuracy with reactivation overhead.