Misc Notes for P3

You do not need to wholescale replace your SkipLists with a B+ tree. There is no need to change your adjacency list. You should use the B+ for your data dictionary and your coordinate checking structure. However, as with the Skiplist the primary benchmark testing will be done by using your B+ tree class with my own code. I happen to think that the naive pointer assignment based B+ should be faster in general than the skiplist, and I hope that for large datasets we can get the assignment based B+ to outperform the skiplist. The easiest and slowest way to do this would be to do grading on a low memory non-cluster machine, but that's probably more work than I'm prepared for ;) I don't think we'll really be able to see the gain unless some of our memory gets paged to disk, and with the amount of memory the cluster machines have available I am not sure that is feasible.

Web Accessibility