B-Trees

Based on materials by D. Frey and

T. Anastasio

1
Large Trees

n  Tailored toward applications where tree

doesn’t fit in memory
q  operations much faster than disk accesses
q  want to limit levels of tree (because each new
level requires a disk access)
q  keep root and top level in memory

2
An Alternative to BSTs

n  Up until now we assumed that each node in a BST

stored the data.
n  What about having the data stored only in the
leaves?
n  The internal nodes just guide our search to the leaf
which contains the data we want.
n  We’ll restrict this discussion of such trees to those in
which all leaves are at the same level.

3
 20

12 40

8 17 33 45

1
9 12 18 27 33
2 40 45
10 15 19 29 37
5 41
20
7
Figure 1 - A BST with data stored in the leaves
4
 Observations

n  Store data only at leaves; all leaves at same level

q  interior and exterior nodes have different structure

q  interior nodes store one key and two subtree

pointers
q  all search paths have same length: ⎡lg n⎤

(assuming one element per leaf)

q  can store multiple data elements in a leaf

5
M-Way Trees

n  A generalization of the previous BST model

q  each interior node has M subtrees pointers

and M-1 keys

n  the previous BST would be called a “2-way tree” or “M-
way tree of order 2”
q  as M increases, height decreases: ⎡lgM n⎤
(assuming one element per leaf)
q  A perfect M-way tree of height h has Mh leaves

6
An M-Way Tree of Order 3
Figure 2 (next page) shows the same data as figure 1,
stored in an M-way tree of order 3. In this example
M = 3 and h = 2, so the tree can support 9 leaves,
although it contains only 8.
One way to look at the reduced path length with
increasing M is that the number of nodes to be visited
in searching for a leaf is smaller for large M.
We’ll see that when data is stored on the disk, each
node visited requires a disk access, so reducing the
nodes visited is essential.

7
 12 20

5 9 15 18 26

27
1 21
5 10 15 18 30
2 12 24
7 11 16 19 34
4
42
Figure 2 -- An M-Way tree of order 3
8
Searching in an M-Way Tree

n  Different from standard BST search

q  search always terminates at a leaf node
q  might need to scan more than one element at a leaf
q  might need to scan more than one key at an interior node
n  Trade-offs
q  tree height decreases as M increases
q  computation at each node during search increases as M
increases

9
Searching an M-Way Tree
Search (MWayNode v, DataType element, boolean foundIt)
if (v == NULL) return failure;
if (v is a leaf)
search the list of values looking for element
if found, return success otherwise return failure
else (if v is an interior node)
search the keys to find which subtree element is in
recursively search the subtree

10
 Search Algorithm: Traversing the M-way Tree

Everything in this
subtree is smaller than
this key 18 32

10 13 22 28 39

1 13 18 23 32
10 28 39
2 14 24 35
11 30 44
9 16 25 38
In any interior node, find the first key > search item, and traverse the link to the left of that key. Search for any
item >= the last key in the subtree pointed to by the rightmost link. Continue until search reaches a leaf.

11
 22 36 48

6 12 18 26 32 42 54

2 6 14 18 22 26 32 38 42 48 54
4 8 16 19 24 28 34 40 44 50 56
10 20 30 46 52

Figure 3 – searching in an M-way tree of order 4

12
Is it worth it?

n  Is it worthwhile to reduce the height of the

search tree by letting M increase?
n  Although the number of nodes visited
decreases, the amount of computation at
each node increases.
n  Where’s the payoff?

13
An example
n  Consider storing 107 items in a balanced
BST and in an M-way tree of order 10.
7
n  The height of the BST will be lg(10 ) ~ 24.

n  The height of the M-Way tree will be

log(107 ) = 7 (assuming that we store just
1 record per leaf)
n  However, in the BST, just one comparison
will be done at each interior node, but in
the M-Way tree, 9 will be done
(worst case)
14
How can this be worth the price?

n  Only if it somehow takes longer to descend the tree

than it does to do the extra computation
n  This is exactly the situation when the nodes are
stored externally (e.g. on disk)
n  Compared to disk access time, the time for extra
computation is insignificant
n  We can reduce the number of accesses by sizing
the M-way tree to match the disk block and record
size.

15
 A Generic M-Way Tree Node
public class MwayNode<Ktype, Dtype>
{
// code for public interface here
// constructors, accessors, mutators

private boolean isLeaf; // true if node is a leaf

private int m; // the “order” of the node
private int nKeys; // nr of actual keys used
private ArrayList<Ktype> keys; // array of keys(size = m - 1)
private MWayNode subtrees[ ]; // array of pts (size = m)
private int nElems; // nr poss. elements in leaf
private List<Dtype> data; // data storage if leaf
}

16
B-Tree Definition

A B-Tree of order M is an M-Way tree with the following

constraints
1.  The root is either a leaf or has between 2 and M subtrees
2.  All interior node (except maybe the root) have between
3.  ⎡M / 2⎤ and M subtrees
(i.e. each interior node is at least “half full”)
4.  All leaves are at the same level. A leaf must store between
⎡L / 2⎤ and L data elements, where L is a fixed
constant >= 1 (i.e. each leaf is at least “half full”,
except when the tree has fewer than L/2 elements)

17
A B-Tree example

n  The following figure (also figure 3) shows a B-Tree

with M = 4 and L = 3
n  The root node can have between 2 and M = 4
subtrees
n  Each other interior node can have between
⎡ M / 2⎤ = ⎡ 4 / 2⎤ = 2 and M = 4 subtrees and up to
M – 1 = 3 keys.
n  Each exterior node (leaf) can hold between
⎡ L / 2⎤ = ⎡ 3 / 2⎤ = 2 and L = 3 data elements

18
 22 36 48

6 12 18 26 32 42 54

2 6 14 18 22 26 32 38 42 48 54
4 8 16 19 24 28 34 40 44 50 56
10 20 30 46 52

Figure 4 – A B-Tree with M = 4 and L = 3

19
Designing a B-Tree

n  Recall that M-way trees (and therefore B-trees) are

often used when there is too much data to fit in
memory. Therefore each node and leaf access
costs one disk access.
n  When designing a B-Tree (choosing the values of M
and L), we need to consider the size of the data
stored in the leaves, the size of the keys and
pointers stored in the interior nodes, and the size of
a disk block.

20
Student Record Example

Suppose our B-Tree stores student records which

contain name, address, etc. and other data totaling
1024 bytes.
Further assume that the key to each student record
(ssn??) is 8 bytes long.
Assume also that a pointer (really a disk block number,
not a memory address) requires 4 bytes
And finally, assume that our disk block is 4096 bytes

21
Calculating L

L is the number of data records that can be

stored in each leaf. Since we want to do just
one disk access per leaf, this is the same as
the number of data records per disk block.
Since a disk block is 4096 and a data record is
1024, we choose L = ⎣4096 / 1024⎦ = 4 data
records per leaf.

22
Calculating M

Each interior node contains M pointers and M-1

keys. To maximize M (and therefore keep the tree
flat and wide) and yet do just one disk access per
interior node, we have the following relationship
4M + 8 ( M – 1) <= 4096
12M <= 4104
M <= 342
So choose the largest possible M (making tree as
shallow as possible) of 342.

23
Performance of our B-Tree
With M = 342 the height of our tree for N students
will be ⎡ log342 ⎡ N/L ⎤ ⎤ .
For example, with N = 100,000 (about 10 times the
size of UMBC student population) the height of
the tree with M = 342 would be no more than 2,
because ⎡ log342(25000)⎤ = 2.
So any student record can be found in 3 disk
accesses. If the root of the B-Tree is stored in
memory, then only 2 disk accesses are needed .

24
Insertion of X in a B-Tree
n  Search to find the leaf into which X should be
inserted
n  If the leaf has room (fewer than L elements), insert X
and write the leaf back to the disk.
n  If the is leaf full, split it into two leaves, each with half
of elements. Insert X into the appropriate new leaf
and write new leaves back to the disk.
q  Update the keys in the parent
q  If the parent node is already full, split it in the same manner
q  Splits may propagate all the way to the root, in which case,
the root is split (this is how the tree grows in height)

25
Insert 33 into this B-Tree
22 36 48

6 12 18 26 32 42 54

2 6 12 18 22 26 32 36 42 48 54
4 8 14 19 24 28 34 38 44 50 56
10 16 20 30 40 46 52

Figure 5 – before inserting 33

26
Inserting 33

n  Traversing the tree from the root, we find that

33 is less than 36 and is greater than 22,
leading us to the 2nd subtree. Since 33 is
greater than 32 we are led to the 3rd leaf (the
one containing 32 and 34).
n  Since there is room for an additional data
item in the leaf it is inserted (in sorted order
which means reorganizing the leaf)

27
 After inserting 33
22 36 48

6 12 18 26 32 42 54

2 6 12 18 22 26 32 36 42 48 54
4 8 14 19 24 28 33 38 44 50 56
10 16 20 30 34 40 46 52

Figure 6 – after inserting 33

28
Now insert 35

n  This item also belongs in the 3rd leaf of the

2nd subtree. However, that leaf is full.
n  Split the leaf in two and update the parent to
get the tree in figure 7.

29
 After inserting 35
22 36 48

6 12 18 26 32 34 42 54

2 6 12 18 22 26 32 34 36 42 48 54
4 8 14 19 24 28 33 35 38 44 50 56
10 16 20 30 40 46 52

Figure 7 – after inserting 35

30
Inserting 21

n  This item belongs in the 4th leaf of the 1st subtree
(the leaf containing 18, 19, 20).
n  Since the leaf is full, we split it and update the keys
in the parent.
n  However, the parent is also full, so it must be split
and its parent (the root) updated.
n  But this would give the root 5 subtrees which is not
allowed, so the root must also be split.
n  This is the only way the tree grows in height

31
 After inserting 21
36

18 22 48

6 12 20 26 32 34 42 54

2 6 12 18 20 22 26 32 34 36 42 48 54
4 8 14 19 21 24 28 33 35 38 44 50 56
10 16 30 40 46 52

Figure 8 – after inserting 21

32
 B-tree Deletion
•  Find leaf containing element to be deleted.
•  If that leaf is still full enough (still has ⎡ L / 2⎤
elements after remove) write it back to disk
without that element. Then change the key in the
ancestor if necessary.
•  If leaf is now too empty (has less than ⎡ L / 2⎤
elements), borrow an element from a neighbor.
•  If neighbor would be too empty, combine two leaves into one.
•  This combining requires updating the parent which may now have too few subtrees.
•  If necessary, continue the combining up the tree
•  Does it matter which neighbor we borrow from?

33