HCloud, a Healthcare-Oriented Cloud System with Improved Efficiency in Biomedical Data Processing - Cloud Computing with e-Science Applications - page 170

8.2.3 Functionalities of Components in HCloud
As mentioned, HCloud is planned in the near future to serve hundreds of
city-dwelling families via Internet health monitoring and will support thou-
sands of potential customers, including physicians and whoever cares about
their health. So, a higher number of concurrent transactions and shorter
response time are requirements of our system, as is large-scale graphic
drawing for visualized results, which need to be stored and accessed effi-
ciently to provide instant services to individuals.
Therefore, the six-layer architecture of the private HCloud platform is
proposed with the philosophy of inheriting the software as a service (SaaS)
of the NIST model and introducing the in-source/outsource concept into
development. Each layer's content and function are interpreted as service
interaction, service presentation, session cache, cloud engine, medical data
mining, and cloud storage, respectively, as shown in Figure 8.3.
• Service interaction is a top layer; users can interoperate with the ter-
minals, such as 3G mobile phone, set-top box, and browser on a com-
puter, to collect and upload original physiological data as well as
download analysis results from the cloud servers.
• Service presentation can be regarded as an interface with various
kinds of services, such as the wireless application protocol (WAP),
web, or image provider. A load balance mechanism is introduced
to the system on this layer so that the web/WAP servers can work
cooperatively as a cluster to maintain optimal performance. The
server cluster shares users' requests to the website together to meet
the high concurrence requirement of health care services and to
better ensure quality of service (QoS). In addition, some runtime
information is reserved at this level.
• The session cache stores the user's sessions on one hand, which main-
tains the authentication and certification status the first time when
the user accesses the services (i.e., information of service interaction
through the presentation layer). To share the sessions among all
web/WAP servers, this platform uses a separate session server to
save the session data, which can solve the problem of session status
sharing when there is load balancing. On the other hand, a memory
cache is adopted to expedite access to the results data. Cache serv-
ers are specialized servers used to save users' pages, documents,
profiles, and so on temporarily. This kind of server can reduce the
capacity of network exchange because displaying graphs of physi-
ological signals would take a long time. This platform will draw
the images when users access the service the first time and save
the graphs in the graph servers while registering the session keys
in the cache server. If the users want to browse a previously gener-
ated image, the image will be loaded from the graph servers directly