Databases Reference
In-Depth Information
Fragment Part1
PartNum
Description
OnHand
Class
Warehouse
Price
CD52
Microwave Oven
32
AP
1
$165.00
KL62
Dryer
12
AP
1
$349.95
Fragment Part2
PartNum
Description
OnHand
Class
Warehouse
Price
BV06
Home Gym
45
SG
2
$794.95
DR93
Gas Range
8
AP
2
$495.00
KV29
Treadmill
9
SG
2
$1,390.00
Fragment Part3
279
PartNum
Description
OnHand
Class
Warehouse
Price
AT94
Iron
50
HW
3
$24.95
DL71
Cordless Drill
21
HW
3
$129.95
DW11
Washer
12
AP
3
$399.99
FD21
Stand Mixer
22
HW
3
$159.95
KT03
Dishwasher
8
AP
3
$595.00
FIGURE 9-3
Fragmentation of Part table data by warehouse
You assign Fragment Part1 to the database at warehouse 1, Fragment Part2 to the database at warehouse 2,
and Fragment Part3 to the database at warehouse 3. The effect of these assignments is that data about each part is
stored in the database at the warehouse where the part is stored. You can access the complete Part table by tak-
ing the union of the three fragments.
Users should not be aware of the fragmentation—they should feel as if they are using a single central
database. When users are unaware of fragmentation, the DDBMS has
fragmentation transparency
.
When compared with a single centralized database, distributed databases offer the following advantages:
Local control of data.
Because each location retains its own data, a location can exercise greater
control over that data. With a single centralized database, on the other hand, the central site that
maintains the database is usually unaware of all the local issues at the various sites served by the
database.
●
Increasing database capacity.
In a properly designed and installed distributed database, the pro-
cess of increasing system capacity is often simpler than in a centralized database. If the size of
the disk at a single site becomes inadequate for its database, you need to increase the capacity of
the disk only at that site. Furthermore, you can increase the capacity of the entire database sim-
ply by adding a new site.
●
System availability.
When a centralized database becomes unavailable for any reason,
no
users
can continue processing. In contrast, if one of the local databases in a distributed database
becomes unavailable, only users who need data in that particular database are affected; other users
can continue processing in a normal fashion. In addition, if the data has been replicated (another
copy of it exists in other local databases), potentially all users can continue processing. How-
ever, processing for users at the site of the unavailable database will be much less efficient because
data that was formerly obtained locally must now be obtained through communication with a
remote site.
●
Improved performance.
When data is available locally, you eliminate network communication
delays and can retrieve data faster than with a remote centralized database.
●
