2019-05-03 in class

notes.md

@@ -313,4 +313,38 @@ k <= a     a <= k <= b     k >= b
 4. If a node is red, both of its children are black
 5. The black height is the same for every branch in the tree
 
-A red node is a way to strech the tree, but the
+A red node is a way to strech the tree, but you cannot stretch it too far.
+
+# B-trees
+
+Disk access sucks (a billion times slower than registers). So, ignore the RAM
+model and come up with a better solution.
+
+If the number of keys per node is increased to k keys, the complexity of tree
+search will be log_k+1(n). We can search between the keys linearly because we
+have to wait for the disks. And log_1000() is much better than log_2().
+
+## Implementation
+
+- Minimum degree (min. num. of children in each node) defined as `t` (`t >= 2`)
+- Every node other than the root must have at least `t - 1` keys
+- Every node must contain at most `2t - 1` keys
+- Every node has a boolean flag for leaf
+
+### Insertion
+
+- Read node from disk
+- If full, split
+- If leaf, insert in the right position
+- If not leaf, recurse on the right branch based on the keys
+
+### Search
+
+- Read node from disk
+- Linearly search between keys
+  - If key found, return True
+  - otherwise, if leaf, return False
+  - otherwise, if not leaf, recurse on the right branch
+
+
+