Database Reference
In-Depth Information
web application. The assumption can be made that the access is restricted due to
a firewall in Google App Engine. When thousands of requests go into Google App
Engine concurrently from the same IP, the firewall may be triggered. Upon some
analysis of how App Engine manages incoming requests by using a HTTP traffic
monitor, it is reasonable to conclude that this may be a security feature around to
prevent denial of service attacks. There seems no way to get around of it, except
reducing the number of requests.
4.4
Discussion
An empirical experiment was carried out to examine the effectiveness of CARE
when applied to testing different cloud platforms. Results indicate CARE is a
feasible approach by directly comparing three major cloud platforms, including
cloud hosting servers and cloud databases. Analysis revealed the importance of
acknowledging different service models, and that the scalability of cloud hosting
servers is achieved in different ways. Horizontal scalability is available to some
extent in Google App Engine, but is always restricted by the quota limitation.
On the contrary, Amazon EC2 and Microsoft Windows Azure can only scale
through manual work in which developers can specify rules and conditions for when
instances should be added. This leads the classic trading off issue of complexity
against scalability. Vertical scalability is not possible in Google App Engine since
every process has to be finished within 30 s, and that it is beyond the control over
the type of machines used for our application in the Google cloud. Where on the
other hand, Amazon EC2 and Microsoft Windows Azure allow you to choose and
deploy instances with varying sizes of memory and CPUs.
The unpredictable unavailability of cloud is of a greater issue, particularly
for enterprise organizations with mission critical application requirements. Whilst
bursts of unavailability are noticed, during the tests which are caused by a range of
environmental factors, including variable network conditions. It is also observed that
the cloud providers sometimes experience challenges in maintaining uninterrupted
service availability. Despite sophisticated replication strategies, there is still a
potential risk of data center breakdown even in the cloud, which may in turn affect
the performance and availability of hosted applications. It is also noticed that at the
time of writing, most cloud vendors provide an SLA availability of 99:9%, which is
still some way away from the typical enterprise requirement of 99:999%.
The network condition makes a significant impact on the total performance
and end-user experience for cloud computing. The performance of the end-to-end
cloud experience highly relies on the network condition. If an end-user accesses
cloud services through a poor network environment, it is not possible to take full
advantage of the cloud platforms.
