Database Reference
In-Depth Information
3.5
Virtualized Database Servers
NoSQL database as a service and relational database as a service offered by cloud
providers both come with their own strengths. Firstly, the customers do not have
to trouble themselves with administrative work, as the providers deal with software
upgrades and maintenance tasks. Secondly, the cloud providers also implemented
automatic replication failover and management. But there are obvious shortcomings
as well. Firstly, customers may require extra migration efforts on modifying code
and converting data. Secondly, customers have limited choices, if customers use
PostgreSQL or DB2 as their database, there is no simple alternative for both
solutions. And thirdly, customers have no full control on achieving the elasticity
and scalability benefits.
Therefore, an approach like
virtualized database servers
is necessary sometimes.
For this approach, customers simply port everything designed for a conventional
data center into cloud, including database servers, and run in virtual machines. It is
worth mentioning that there is no unique approach of deploying virtualized database
servers. Therefore, no specific projects and examples will be discussed in this
subsection. The virtualized database servers are considered as being good enough,
as long as the deployment meets the application requirements.
With such a deployment, there would be minimum changes to existing applica-
tion code. The customers have full control in configuring the required elasticity of
allocated resources [
96
,
211
]. And the customers can also build low cost solutions
for geographic replication by taking advantage of cloud providers' multiple data
centers across continents. However, achieving these goals requires the existence of
control components [
207
] which are responsible for monitoring the system state and
taking the corresponding actions, such as allocating more/less computing resources,
according to the defined application requirements and strategies. Several approaches
have been proposed for building control components which are based on the
efficiency of utilization of the allocated resources [
96
,
211
]. In Chap.
7
, we present
our proposed approach that focuses on building an SLA-based admission control
component that provides a customer-centric view for achieving the requirements of
their applications.
3.6
Web Scale Data Management: Trade-Offs
An important issue in designing large scale data management applications is to
avoid the mistake of trying to be “everything for everyone”. As with many types of
computer systems, no one system can be best for all workloads and different systems
make different tradeoffs in order to optimize for different applications. Therefore,
the most challenging aspects in these application is to identify the most important
features of the target application domain and to decide about the various design
trade-offs which immediately lead to performance trade-offs. To tackle this problem,
