Database Reference
In-Depth Information
3 Autonomic Deployment Operations
By studying the state-of-the-art on IoT solutions, it is seen that generally IoT platforms
consists of different modules developed to support a business model approach (vertical
approach) [
4
,
5
]. A vertical approach seeks for a single stack organisation in order to
simplify operations and generate sort of abstraction that eventually can
t into a
business model. At IoT verticals isolated and almost cero-exchange of information can
get from them however a more controlled environment(s) can be offered as guarantee
for better management and system stability. As consequence of this trade-off between
control and stability system developers has adopted vertical designs instead of hori-
zontal, building in this form IoT silos.
In the other hand, as part of the main features in IoT systems, it is required
operations and services are inter-connected and inter-dependent. The service modules
have dependencies in both functional and non-functional level. Particularly the pro-
posed architecture has been developed keeping in mind the service computing para-
digm, and the reusability and self-con
guration of the platform itself (autonomics). The
deployment platform might change, the requirements of the test-bed might change, and
the dependencies might be different than the initial set-up. An autonomic design will
support these and other self-x features. In practical terms an autonomic deployment will
create a virtualization image in front of the actual IoT test-bed and provide and esti-
mate, before hand, with the details that are necessary to deploy the system. Although
lots of work has been done in the area of autonomics in IoT test-bed [
6
,
7
] and
distributed systems [
8
11
], Sensor Network Management [
12
], Dynamic Deployment
of Software Component [
13
-
15
]. The Autonomic Deployment Agent for IoT test-bed
has been designed following the IoT Autonomic lifecycle. The following identi
-
ed
autonomic operations for IoT Service Delivery model are parts of the requirements for
designing the Autonomic architecture explained in the Sect.
4
.
3.1 Data Collection and Sensor Monitoring
During the data collection and monitoring phase, the agent analyses the system and
nd out what resources are available, and also the requirements (e.g.: dependencies) of
the IoT platform for deployment. It will map these two variables and will hold a
deployment_pair in store to match the best combination with future pairs.
3.2 Analysis and Transformations
During this phase, the agent de
nes ways to resolve the dependencies. Some depen-
dencies might be internal, for example one module depends on the deployment of
another, and some will be external where the required resource has to be extracted from
an external source. A dependency map is created in this phase and the system creates
the list of dependencies (linked data) that needs to be resolved
during the
Execution phase. In the Dependency map, the agent analyses and determine whether
the dependencies can be resolved internally or it has to be dealt and solved with
external changes such as increment of cluster memory or adding a new processor.
“
somehow
”
Search WWH ::
Custom Search