Outline

 Basic Concepts
 Ordered Indices
 B+-Tree Index Files
 B-Tree Index Files
 Hashing
 Write-optimized indices
 Spatio-Temporal Indexing

1/18/2021 1
 Basic Concepts

 Indexing mechanisms used to speed up access to desired data.

• E.g., author catalog in library
 Search Key - attribute to set of attributes used to look up records in a file.
 An index file consists of records (called index entries) of the form

search-key pointer
 Index files are typically much smaller than the original file
 Two basic kinds of indices:
• Ordered indices: search keys are stored in sorted order
• Hash indices: search keys are distributed uniformly across “buckets” using a “hash
function”.

1/18/2021 2
 Index Evaluation Metrics

 Access types supported efficiently. E.g.,

• Records with a specified value in the attribute
• Records with an attribute value falling in a specified range of values.
 Access time
 Insertion time
 Deletion time
 Space overhead

1/18/2021 3
 Ordered Indices

 In an ordered index, index entries are stored sorted on the search key value.
 Clustering index: in a sequentially ordered file, the index whose search key specifies the
sequential order of the file.
• Also called primary index
• The search key of a primary index is usually but not necessarily the primary key.
 Secondary index: an index whose search key specifies an order different from the
sequential order of the file. Also called
nonclustering index.
 Index-sequential file: sequential file ordered on a search key, with a clustering index on the
search key.

1/18/2021 4
 Dense Index Files

 Dense index — Index record appears for every search-key value in the file.
 E.g. index on ID attribute of instructor relation

1/18/2021 5
Dense Index Files (Cont.)

 Dense index on dept_name, with instructor file sorted on dept_name

1/18/2021 6
 Sparse Index Files

 Sparse Index: contains index records for only some search-key values.
• Applicable when records are sequentially ordered on search-key
 To locate a record with search-key value K we:
• Find index record with largest search-key value < K
• Search file sequentially starting at the record to which the index record points

1/18/2021 7
 Sparse Index Files (Cont.)

 Compared to dense indices:

• Less space and less maintenance overhead for insertions and deletions.
• Generally slower than dense index for locating records.
 Good tradeoff:
• for clustered index: sparse index with an index entry for every block in file, corresponding to least
search-key value in the block.

• For unclustered index: sparse index on top of dense index (multilevel index)

1/18/2021 8
 Secondary Indices Example

 Secondary index on salary field of instructor

 Index record points to a bucket that contains pointers to all the actual records with that particular search-
key value.
 Secondary indices have to be dense

1/18/2021 9
 Multilevel Index

 If index does not fit in memory, access becomes expensive.

 Solution: treat index kept on disk as a sequential file and construct a sparse index on
it.
• outer index – a sparse index of the basic index
• inner index – the basic index file
 If even outer index is too large to fit in main memory, yet another level of index can be
created, and so on.
 Indices at all levels must be updated on insertion or deletion from the file.

1/18/2021 11
Multilevel Index (Cont.)

1/18/2021 12
Indices on Multiple Keys

 Composite search key

• E.g., index on instructor relation on attributes (name, ID)
• Values are sorted lexicographically
 E.g. (John, 12121) < (John, 13514) and
(John, 13514) < (Peter, 11223)
• Can query on just name, or on (name, ID)

1/18/2021 15
Example of B+-Tree

1/18/2021 17
 B+-Tree Index Files (Cont.)

A B+-tree is a rooted tree satisfying the following properties:

 All paths from root to leaf are of the same length
 Each node that is not a root or a leaf has between n/2 and n children.
 A leaf node has between (n–1)/2 and n–1 values
 Special cases:
• If the root is not a leaf, it has at least 2 children.
• If the root is a leaf (that is, there are no other nodes in the tree), it can have between 0
and (n–1) values.

1/18/2021 18
 B+-Tree Node Structure

 Typical node

• Ki are the search-key values

• Pi are pointers to children (for non-leaf nodes) or pointers to records or buckets of records
(for leaf nodes).
 The search-keys in a node are ordered
K1 < K2 < K3 < . . . < Kn–1
(Initially assume no duplicate keys, address duplicates later)

1/18/2021 19
 Leaf Nodes in B+-Trees

Properties of a leaf node:

 For i = 1, 2, . . ., n–1, pointer Pi points to a file record with search-key value Ki,
 If Li, Lj are leaf nodes and i < j, Li’s search-key values are less than or equal to Lj’s search-key values
 Pn points to next leaf node in search-key order

1/18/2021 20
 Non-Leaf Nodes in B+-Trees

 Non leaf nodes form a multi-level sparse index on the leaf nodes. For a non-leaf node with m pointers:
• All the search-keys in the subtree to which P1 points are less than K1
• For 2  i  n – 1, all the search-keys in the subtree to which Pi points have values greater than or equal to Ki–1 and
less than Ki
• All the search-keys in the subtree to which Pn points have values greater than or equal to Kn–1
• General structure

1/18/2021 21
 Example of B+-tree

 B+-tree for instructor file (n = 6)

 Leaf nodes must have between 3 and 5 values

((n–1)/2 and n –1, with n = 6).
 Non-leaf nodes other than root must have between 3 and 6 children ((n/2 and n with n =6).
 Root must have at least 2 children.

1/18/2021 22
 Observations about B+-trees

 Since the inter-node connections are done by pointers, “logically” close blocks need not be “physically” close.
 The non-leaf levels of the B+-tree form a hierarchy of sparse indices.
 The B+-tree contains a relatively small number of levels
 Level below root has at least 2* n/2 values
 Next level has at least 2* n/2 * n/2 values
 .. etc.
• If there are K search-key values in the file, the tree height is no more than  logn/2(K)
• thus searches can be conducted efficiently.
 Insertions and deletions to the main file can be handled efficiently, as the index can be restructured in logarithmic time (as
we shall see).

1/18/2021 23
 Queries on B+-Trees

function find(v)
1. C=root
2. while (C is not a leaf node)
1. Let i be least number s.t. V  Ki.
2. if there is no such number i then
3. Set C = last non-null pointer in C
4. else if (v = C.Ki ) Set C = Pi +1
5. else set C = C.Pi
3. if for some i, Ki = V then return C.Pi
4. else return null /* no record with search-key value v exists. */

1/18/2021 24
Queries on B+-Trees (Cont.)

 Range queries find all records with search key values in a given range
• See book for details of function findRange(lb, ub) which returns set of all such records
• Real implementations usually provide an iterator interface to fetch matching records one at a time, using a next()
function

1/18/2021 25
 Queries on B+-Trees (Cont.)

 If there are K search-key values in the file, the height of the tree is no more than logn/2(K).
 A node is generally the same size as a disk block, typically 4 kilobytes
• and n is typically around 100 (40 bytes per index entry).
 With 1 million search key values and n = 100
• at most log50(1,000,000) = 4 nodes are accessed in a lookup traversal from root to leaf.
 Contrast this with a balanced binary tree with 1 million search key values — around 20 nodes are accessed in a lookup
• above difference is significant since every node access may need a disk I/O, costing around 20 milliseconds

1/18/2021 26
Non-Unique Keys

 If a search key ai is not unique, create instead an index on a composite key (ai , Ap), which is unique
• Ap could be a primary key, record ID, or any other attribute that guarantees uniqueness
 Search for ai = v can be implemented by a range search on composite key, with range (v, - ∞) to (v, + ∞)
 But more I/O operations are needed to fetch the actual records
• If the index is clustering, all accesses are sequential
• If the index is non-clustering, each record access may need an I/O operation

1/18/2021 27
Updates on B+-Trees: Insertion

Assume record already added to the file. Let

 pr be pointer to the record, and let
 v be the search key value of the record
1. Find the leaf node in which the search-key value would appear
1. If there is room in the leaf node, insert (v, pr) pair in the leaf node
2. Otherwise, split the node (along with the new (v, pr) entry) as discussed in the next slide, and propagate updates to
parent nodes.

1/18/2021 28
Updates on B+-Trees: Insertion (Cont.)

 Splitting a leaf node:

• take the n (search-key value, pointer) pairs (including the one being inserted) in sorted order. Place the first n/2 in
the original node, and the rest in a new node.
• let the new node be p, and let k be the least key value in p. Insert (k,p) in the parent of the node being split.
• If the parent is full, split it and propagate the split further up.
 Splitting of nodes proceeds upwards till a node that is not full is found.
• In the worst case the root node may be split increasing the height of the tree by 1.

Result of splitting node containing Brandt, Califieri and Crick on inserting Adams
Next step: insert entry with (Califieri, pointer-to-new-node) into parent

1/18/2021 29
 B+-Tree Insertion

Affected nodes

B+-Tree before and after insertion of “Adams”

1/18/2021 30
B+-Tree Insertion

B+-Tree before and after insertion of “Lamport”

Affected nodes

1/18/2021 Affected nodes 31

Insertion in B+-Trees (Cont.)

 Splitting a non-leaf node: when inserting (k,p) into an already full internal node N
• Copy N to an in-memory area M with space for n+1 pointers and n keys
• Insert (k,p) into M
• Copy P1,K1, …, K n/2-1,P n/2 from M back into node N
• Copy Pn/2+1,K n/2+1,…,Kn,Pn+1 from M into newly allocated node N'
• Insert (K n/2,N') into parent N
 Example

 Read pseudocode in book!

1/18/2021 32
 Examples of B+-Tree Deletion

Before and after deleting “Srinivasan”

Affected nodes

 Deleting “Srinivasan” causes merging of under-full leaves

1/18/2021 33
Examples of B+-Tree Deletion (Cont.)

Before and after deleting “Singh” and “Wu”

Affected nodes

 Leaf containing Singh and Wu became underfull, and borrowed a value Kim from its left sibling
 Search-key value in the parent changes as a result

1/18/2021 34
 Example of B+-tree Deletion (Cont.)

Before and after deletion of “Gold”

 Node with Gold and Katz became underfull, and was merged with its sibling
 Parent node becomes underfull, and is merged with its sibling
• Value separating two nodes (at the parent) is pulled down when merging
 Root node then has only one child, and is deleted

1/18/2021 35
Updates on B+-Trees: Deletion

Assume record already deleted from file. Let V be the search key value of the record, and Pr be the pointer to the record.
 Remove (Pr, V) from the leaf node
 If the node has too few entries due to the removal, and the entries in the node and a sibling fit into a single node, then
merge siblings:
• Insert all the search-key values in the two nodes into a single node (the one on the left), and delete the other node.
• Delete the pair (Ki–1, Pi), where Pi is the pointer to the deleted node, from its parent, recursively using the above
procedure.

1/18/2021 36
Updates on B+-Trees: Deletion

 Otherwise, if the node has too few entries due to the removal, but the entries in the node and a sibling do not fit into a
single node, then redistribute pointers:
• Redistribute the pointers between the node and a sibling such that both have more than the minimum number of
entries.
• Update the corresponding search-key value in the parent of the node.
 The node deletions may cascade upwards till a node which has n/2 or more pointers is found.
 If the root node has only one pointer after deletion, it is deleted and the sole child becomes the root.

1/18/2021 37
Complexity of Updates

 Cost (in terms of number of I/O operations) of insertion and deletion of a single entry proportional to height of the tree
• With K entries and maximum fanout of n, worst case complexity of insert/delete of an entry is O(logn/2(K))
 In practice, number of I/O operations is less:
• Internal nodes tend to be in buffer
• Splits/merges are rare, most insert/delete operations only affect a leaf node
 Average node occupancy depends on insertion order
• 2/3rds with random, ½ with insertion in sorted order

1/18/2021 38
Non-Unique Search Keys

 Alternatives to scheme described earlier

• Buckets on separate block (bad idea)
• List of tuple pointers with each key
 Extra code to handle long lists
 Deletion of a tuple can be expensive if there are many duplicates on search key (why?)
• Worst case complexity may be linear!
 Low space overhead, no extra cost for queries
• Make search key unique by adding a record-identifier
 Extra storage overhead for keys
 Simpler code for insertion/deletion
 Widely used

1/18/2021 39
B+-Tree File Organization

 B+-Tree File Organization:

• Leaf nodes in a B+-tree file organization store records, instead of pointers
• Helps keep data records clustered even when there are insertions/deletions/updates
 Leaf nodes are still required to be half full
• Since records are larger than pointers, the maximum number of records that can be stored in a leaf node is less than
the number of pointers in a nonleaf node.
 Insertion and deletion are handled in the same way as insertion and deletion of entries in a B +-tree index.

1/18/2021 40
B+-Tree File Organization (Cont.)

 Example of B+-tree File Organization

 Good space utilization important since records use more space than pointers.
 To improve space utilization, involve more sibling nodes in redistribution during splits and merges
• Involving 2 siblings in redistribution (to avoid split / merge where possible) results in each node having at least
entries

 2n / 3

1/18/2021 41
Other Issues in Indexing

 Record relocation and secondary indices

• If a record moves, all secondary indices that store record pointers have to be updated
• Node splits in B+-tree file organizations become very expensive
• Solution: use search key of B+-tree file organization instead of record pointer in secondary index
 Add record-id if B+-tree file organization search key is non-unique
 Extra traversal of file organization to locate record
• Higher cost for queries, but node splits are cheap

1/18/2021 42
Indexing Strings

 Variable length strings as keys

• Variable fanout
• Use space utilization as criterion for splitting, not number of pointers
 Prefix compression
• Key values at internal nodes can be prefixes of full key
 Keep enough characters to distinguish entries in the subtrees separated by the key value
• E.g., “Silas” and “Silberschatz” can be separated by “Silb”
• Keys in leaf node can be compressed by sharing common prefixes

1/18/2021 43
Bulk Loading and Bottom-Up Build

 Inserting entries one-at-a-time into a B +-tree requires  1 IO per entry

• assuming leaf level does not fit in memory
• can be very inefficient for loading a large number of entries at a time (bulk loading)
 Efficient alternative 1:
• sort entries first (using efficient external-memory sort algorithms discussed later in Section 12.4)
• insert in sorted order
 insertion will go to existing page (or cause a split)
 much improved IO performance, but most leaf nodes half full
 Efficient alternative 2: Bottom-up B+-tree construction
• As before sort entries
• And then create tree layer-by-layer, starting with leaf level
 details as an exercise
• Implemented as part of bulk-load utility by most database systems

1/18/2021 44
 B-Tree Index File Example

B-tree (above) and B+-tree (below) on same data

1/18/2021 47
Indexing on Flash

 Random I/O cost much lower on flash

• 20 to 100 microseconds for read/write
 Writes are not in-place, and (eventually) require a more expensive erase
 Optimum page size therefore much smaller
 Bulk-loading still useful since it minimizes page erases
 Write-optimized tree structures (discussed later) have been adapted to minimize page writes for flash-optimized search
trees

1/18/2021 48
Indexing in Main Memory

 Random access in memory

• Much cheaper than on disk/flash
• But still expensive compared to cache read
• Data structures that make best use of cache preferable
• Binary search for a key value within a large B +-tree node results in many cache misses
 B+- trees with small nodes that fit in cache line are preferable to reduce cache misses
 Key idea: use large node size to optimize disk access, but structure data within a node using a tree with small node size,
instead of using an array.

1/18/2021 49