Information Technology Reference
In-Depth Information
behaviour are effectively still unknown and outline the necessary work that needs to
be done in order to improve future exploitation of cloud systems (section 4).
2
Cloud Delimiting Factors
To really exploit the full potential of cloud environments, it is absolutely necessary to
first understand what clouds are and thereby which capabilities they really offer. Even
though the concepts are widely known, the principles behind their realisation, and
thus their limitations are less well documented. This is due to the quick uptake on the
market, as well as because there is no reference technology for realising clouds,
though Amazon EC2 and Google Docs are the de facto reference infrastructures.
According to the cloud report published by the European Commission, the primary
cloud characteristics are specifically [3]:
•
Utility Computing
•
Elasticity
•
Availability & Reliability
•
Ease of Use
2.1
Size and Interconnect
A cloud environment must thus consist of multiple computing systems that can
dynamically host multiple instances of the same service / application. In other words,
that can replicate the functionality offered according to the current demand, and also
reduce it in a similar fashion. Typically, this is realised by exploiting virtualisation
technologies that host the respective logic, but can be easily encapsulated and
therefore moved between instances, respectively replicated as a full image. The main
point is that this behaviour is transparent to the user (i.e. does not require
reconfiguration of their systems) and that it is steered according to the load,
respectively availability requirements.
Due to technical constraints, elasticity is considerably slow, as distribution of the
image typically involves communication of a large quantity of data, in particular if the
image packs the whole operating system and execution environment of the service in
question. With increasing complexity of the service and its runtime environment, the
according delay implicitly increases and thus makes reaction to availability
requirements slower. This is however a technical constraint posed by the typical
implementation approach, and not by the cloud concepts as such.
As more and more applications move to the cloud, more and more users access
internet-based services and the scale of individual applications increases to
compensate the performance needs, the technical constraints become the major
impeding factors for the fulfilment of the main cloud characteristics. This means that
the number of resources available in a cloud environment may quickly become
insufficient for the needs of the services, respectively users - in particular at times of
peak demand. This leads to the same resource utilization problem again that kicked
off the cloud concepts in the first instance.
Communication limitations become serious impeding factors for performance of
web-based services. Not only the degree of sharing between connections and users,
but also distance between server and client play a major role for this factor. Whilst
