Ashish Kumar

Dept. of CSE
Manipal University Jaipur

1
File Organization
— Even though the database shows us data in the form of
relations we must understand that it is just a logical
representation.
— Finally all the data needs to be stored on the hard disk as
files.
— It is very important in performance point of view that this
organization should allow the database software to access
data quickly and in an efficient way.

2
Files on Disk
— How to store the records on hard disk?
— You can rely on operating systems. But, they way we access
normal data is totally different from the way we access
database data.
— In database, we might not require the whole file, just one or
two records are required. So, we let DBMS use its very own
organization.
— Database ====> Files / Block ====> Record ====> Fields
— We divide each table in block & then store it in hard disk.
— If the size of a block is 1024 byte and 1 record is of size 4
byte then how many records in one block you can save?
— 1024 / 4 =256 i.e; blocking factor is the average number of
record per block.
3
Fixed Length v/s Variable Length Records
— Consider a list of mobile numbers where each number is
exactly 10 digits long.
— Retrieving number will be simple here because we can read
10 characters at a time.
— But consider a list of names. How do we know the size of
names? They can vary.
— To make matters more complex we can have records of
certain number of fixed length entries and some variable
length entries.

4
Records on Disk in Blocks
— Spanned: It allows partial part of record to be stored in a
block.
R1 R2 R3 R4 R4 R5 R6 R7

— Advantages : no wastage of memory.

— Disadvantage: No. of block access will increase to access a
record.
— Unspanned: No record can be stored in more than one
block.
R1 R2 R3 R4 R5 R6 R7 ||||
— Advantages : No. of block access will be less to access a
record.
— Disadvantage: no wastage of memory.
5
Records on Disk in Files
— Ordered File Organization: All records in a file are ordered on
some search key value.
— Searching can be done in binary search mode.
— Advantage: Searching can be efficient. Only when we search
on search key value. If we search on other attribute then no
advantage.
— Disadvantage: Insertion will be expensive due to
reorganization of the entire file.
— Un-Ordered File Organization: All records in a file are inserted
wherever the place is available (usually at the end of file).
— Searching can be done in linear search mode.
— Advantage: Insertion of a record is efficient.
— Disadvantage: Searching is very inefficient. 6
Indexing
— Indexing mechanisms used to speed up access to desired
data. E.g., author catalog in library
— Search Key - attribute or 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 Block -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”.
7
Classification of Indexing
Non-
Primary key + Key/Candidate
Ordered Key + Unordered

Non-Key
+ ordered

These all are single level indexing.

8
Primary Indexing
— Data file is ordered on primary key & we will build index on
primary key.
— A primary index is an ordered file whose records are of
fixed length with two fields. First field is same as primary
key of data file and second field is a pointer to the data
block where key is available.
— Index is created for the first record of each block is known
as block anchors.

9
Dense Indexing
— Dense index — Index record appears for every search-key
value in the file.

10
Sparse Indexing
— Sparse Index: contains index records for only some search-
key values.
— Applicable when records are sequentially ordered on
search-key

11
Secondary Indexing
— Secondary Index provides a
secondary means of accessing
a file for which primary access
already exist.
— It will be dense index. i.e.,
index will be created for every
record in a file.
— Secondary Index does not
have any impact on how the
rows are actually organized in
data blocks.
— They can be in any order. The
only ordering is w.r.t the
index key in index blocks. 12
Clustering Indexing
— It is created on data file whose records are physically
ordered on a non-key attribute which does not have
distinct value for each record.

13
Primary Index Example
— Suppose that we have an ordered file of 30,000 records on a
disk with block size of 1024 bytes. Records are fixed and are
unspanned of size 100 bytes. Suppose we have created
primary index on the key filed of the size 9 bytes and a
block pointer of size 6 bytes, then find the average number
of block access to search a record with and without index.
Without Indexing:
— Record / block = 1024/ 100 = 10.24
— Since it is unspanned, data / block = 10
— Data block required to hold 30,000 = 30,000 / 10 = 3,000
— Block access to search a record = log2 3000 = 11.55 =
Approx. 12
14
Primary Index Example
With Indexing:
— Index record size = 9 + 6 = 15 bytes
— Record / block = 1024/ 15 = 68.266
— Since it is unspanned, data / block = 68
— Since it is primary index, no. of index record = No. of data
block = 3,000 (Due to block anchors)
— Block access to search a record = 3000 / 68 = 44.11 =
Approx. 45
— No. of block access required = log2 45 + 1 = 6 + 1 = 7

15
Secondary Index Example
— Same Ques as above.

Without Indexing:
— Record / block = 1024/ 100 = 10.24
— Since it is unspanned, data / block = 10
— Data block required to hold 30,000 = 30,000 / 10 = 3,000
— Since the data records are unsorted, block access to search
a record = 3000

16
Secondary Index Example
With Indexing:
— Index record size = 9 + 6 = 15 bytes
— Record / block = 1024/ 15 = 68.266
— Since it is unspanned, data / block = 68
— No. of index records = 30,000
— No. of blocks required = 30,000 / 68 = 441.176 = Approx 442
— No. of block access required = log2 442 + 1 = 9 + 1 = 10

17
Thank You

18