Information Technology Reference
In-Depth Information
d
0
d
1
d
2
d
3
r
0
r
1
Parity group
0
v
0
v
1
v
2
v
3
c
0,0
c
0,1
v
4
v
5
v
6
v
7
c
1,0
c
1,1
v
8
v
9
v
10
v
11
c
2,0
c
2,1
v
12
v
13
v
14
v
15
c
3,0
c
3,1
v
16
v
17
v
18
v
19
c
4,0
c
4,1
Figure 15.11
Data placement before addition of a new node
d
0
d
1
d
2
d
3
d
4
r
0
r
1
v
0
v
1
v
2
v
3
v
4
c
'
0,0
c
'
0,1
Parity group
0
v
5
v
6
v
7
v
8
v
9
c
'
1,0
c
'
1,1
c
'
2,0
c
'
2,1
v
10
v
11
v
12
v
13
v
14
v
15
v
16
v
17
v
18
v
19
c
'
3,0
c
'
3,1
Figure 15.12
Data placement after adding one data node
On the other hand, if a central archive server storing a copy of all media data is available,
then it can simply regenerate the new redundant data blocks locally and send them to the
redundant nodes to replace the obsolete redundant data blocks. In this case, the number of
block movements required will be equal to (
B
/
−
h
)). Nevertheless maintaining this central
archive server will incur additional costs and complicatemanagement of the system. Depending
on the particular application this approach may not be cost-effective.
(
N
15.4.2 Reuse of Original Redundant Data
To understand how we can reduce the redundant data update overhead, we first need to study
the computation of the redundant blocks using the RSE code. Let
B
be the total number of
fixed-size media data blocks in the system and denote the
j
th block of a media object by
v
j
.
For simplicity we consider only one media object in the rest of the section. The results can be
readily extended to multiple media objects.
Search WWH ::
Custom Search