Image Processing Reference
In-Depth Information
Table 7.3 Cloud rack terminals characteristics
Device
#
CPU
(GHz)
RAM
(GB)
Core
no.
Hard disk
(GB)
Cache
(MB)
Core speed
(GHz)
Upload speed
(Mbits/sec)
1
2.1
8
Intel Duo
600
2
5
0.6
2
2.3
16
Quad
6600
500
2
5
0.6
3
2.1
4
i5
400
2
3
0.6
4
4
16
i5
1,000
2
5
0.6
5
2.1
8
i7
600
2
3
0.6
6
2.3
16
i5
500
2
5
0.6
7
2.1
4
Quad
6600
400
2
3
0.6
8
4
16
i5
1,000
2
5
0.6
Java-based simulator. The simulated scenario was evaluated for the achievable
throughput, reliability degree, and delay-bounded transmissions (task that has
delay limitations). The mobility model used by the mobile users is based on
probabilistic Fractional Brownian Motion (FBM) [
26
], where nodes are moving,
according to certain probabilities, location, and time. Topology of a “grid” based
network was modeled, according to the grid approach described in [
27
]. Each
MN
m
node directly communicates with other nodes, if the area situated is in the same
(3
3 centre) rectangular area of the node. For the simulation of the proposed
scenario, the varying parameters described in previous section were used,
exploiting a two-dimensional network, consisting of maximum 100
MN
m
/per
r
k
,
(i.e., terminal nodes) located in measured area.
Different sets of experiments were conducted, in order to examine the impact of
the different capacities in the proposed resource migration scheme, as well as
measures for the sizes of the tasks to be migrated. Table
7.3
contains the different
cloud terminals characteristics, regarding the terminal processing power and mem-
ory capabilities, in order to serve as migrated terminals that will host the partitioned
resource up-on the represented cloud rack. Large files are files that are between
20 and 200 Mbytes, whereas small files are files with capacity between the range of
0.1 and 20 Mbytes. The mean service time with the number of racks is also shown in
Fig.
7.4
. The mean service time is greater for large files that are not migrated in
partitionable parts to other terminals on the cloud racks.
The percentage of migrated resources with the number of cloud racks that are
utilized is shown in Fig.
7.5
. It is obvious that large task/application files are
migrated, in order to be served in an offloaded manner. Concurrent claimed
resources with the number of “in-service” cloud racks are shown in Fig.
7.5
.This
figure shows the load of the resources that can be concurrently served through a
cloud rack with the technical characteristics of the terminals shown in Table
7.3
(see Fig.
7.6
). In addition, the Total Service Time in contrast with the technical
characteristics of the particular devices utilized (shown in Table
7.3
) is shown in
Fig.
7.7
. Moreover, Fig.
7.7
presents the complementary cumulative distribution
