Database Reference
In-Depth Information
PRODUCT
relation
CUSTOMER
relation
Storage
Address
10
11
12
13
14
Storage
Address
20
21
22
23
24
CustNo
123
234
345
456
567
CustName
Customer1
Customer2
Customer3
Customer4
Customer5
ProdNo
1
2
3
4
5
ProdDesc
Prod1
Prod2
Prod3
Prod4
Prod5
SALE
relation
Directory with
join-attribute
values
Storage
Address
30
31
32
33
34
SaleDte
2/1/2003
2/15/2003
2/15/2003
2/22/2002
CustNo
345
234
567
123
123
ProdNo
1
3
3
5
4
Units
25
35
40
45
60
CustNo
Value
123
123
234
345
567
Addr to
CUSTOMER
10
10
11
12
14
Addr to
PRODIUCT
24
23
22
20
22
2/24/2002
Figure 12-21
Preconstructed join directory table.
be made against this new table first and then followed by retrievals from the base
tables. Figure 12-21 illustrates this method of performance improvement.
You should use this method for large primary tables where joins on the fly are
expensive in terms of performance overhead. However, if you create too many pre-
constructed join tables, you may be propagating too much data redundancy. Every
time you add new rows to the participating tables, you have to insert new rows in
the directory table. This additional processing may offset any advantage gained with
the preconstructed join table.
CHAPTER SUMMARY
Physical design is the final stage in the design process.
Physical design implements the database on physical storage.
When you make the transition from logical design to physical design, you map
the components of one to those of the other.
Database performance and data management form the primary focus of
physical design.
Physical design components: file organization, indexes, integrity constraints,
data volumes, data usage, and data distribution.
Physical design major tasks: translate global logical model into physical
model,
design representation on physical storage,
and design security
procedures.
Data dictionary or catalog records the components of the design phase.
Data is stored as files in physical storage; a file consists of data blocks; a block
contains records; typically, a record is a row of a relational table.
Search WWH ::




Custom Search