Distributed Systems - Guide to Cloud Computing for Business and Technology Managers - page 75

3.1.4.2 Application Management
Application management refers to the tasks necessary to set up the appli-
cation and to keep the application running successfully and includes the
following:
• Coniguration : The ability to set up the system and assign resources,
including specifications of the number of servers, the number of
clients supported concurrently, assignment of servers to particular
machines, and assignment of components to servers. For high avail-
ability, the configuration is changeable dynamically while the sys-
tem is in operation.
• Monitoring : The ability to get statistics about the system's con-
figuration and operation, including the number of connected
users, the status of servers, resources utilization of servers, and
throughput and response times. For high availability, tracing, and
troubleshooting, systems automatically monitor information such
as server status and also respond to detected errors or predefined
events.
• Controlling : The ability to effect system configuration and operations,
including the ability to start and stop servers or components and
modify parameters.
3.1.4.3 Application Deployment
Application deployment relates to the issues of initial deployment, securing
the application and upgrading it. The ideal deployment scenario involves no
special software on the client at all, as is the case with browser-based thin
clients.
The application architecture must support application versions and the
ability for multiple versions of the same application to operate simultane-
ously. Security is also a critical requirement for the deployment of an appli-
cation (see Sections 19.1.2 through 19.1.4).
3.2 Component-Based Technologies
In traditional procedural programming, the attributes (data) and the behav-
ior (operations on data) are usually separated. Since there is no higher-level
grouping (e.g., object) that binds the attributes and the behavior, a devel-
oper can never really know the entire result of a procedure call and therefore
never knows exactly all the data that are affected by that procedure call. The
detachment of the procedure from the data means that several unrelated