Diploma Thesis Santiago Gómez Sáez - IAAS

3. Related Works
store may contain one or more users. In our prototype we ensure tenant and user isolation at
both communication, and administration and management levels.
Over the past decades, caching has become the key technology in bridging the performance
gap across memory hierarchies via temporal or spatial localities; in particular, the effect is
prominent in disk storage systems [HLS + 12]. Han et al. investigate how cost efficiency in
a Cloud environment can be achieved, specially in applications which require a high I/O
activities number, and present a CaaS (cache-as-a-service) model. Cloud providers offering
data storage solutions present pricing models based on the storage size, usage per time,
or number of requests. Amazon RDS costs $0.025 per hour for a Micro DB Instance usage
[Amac], while Amazon DynamoDB $0.01 per hour for every 50 units of read capacity [Amaa],
and Google Cloud Storage $0.01 per 1000 PUT, POST, GET requests per month [Gooc]. An
I/O-intensive application whose database is hosted in the Cloud may produce a significant
economic cost. The cost of continuously retrieving data from the Cloud data store, when
existing temporal proximity between the data accessed, can be considered unnecessary, and
reducible. Furthermore, the application’s overall performance can be reduced due to the
network latency and, in the scope of this work, the use of an ESB to access the Cloud data store.
In this diploma thesis we do not provide cashing as a service, but include cashing support
to the sharded-based router pattern described in [SAB + 12]. Uralov enhaces ServiceMix-mt
with cashing support for dynamic discovery and selection of Cloud data hosting solutions
[Ura12]. However, we must adapt and extend it due to the lack of support of functionalities
we require and the lack of full OSGi compliance.
3.1. SQL Approaches
In this section we discuss about the different approaches ESB vendors take into account
when accessing or storing data in SQL database systems, and compare it to the support
we provide in this work. Most of the ESB vendors provide support for data persistency
in RDBMS. However, it is restricted to output connections to DBMS. Fuse ESB provides
JDBC data source support, enabling users to connect to a database and make SQL based
queries and updates [Fus11]. This connection can be established during routing, when
receiving a JMS or HTTP message in a consumer endpoint, etc. However, they do not provide
support for native database communication protocol incoming messages, e.g. MySQL or
PostgreSQL communication protocol. Thus, data consumer endpoints supporting native
database protocols are not deployable. The same limited support is provided in the integration
framework Camel, in its component Camel-jdbc [Thec].
JBoss presents its Enterprise Data Service Platform containing data services with SOA support,
and an ESB [Red11]. Any organization currently using an ESB can interoperate their Data
Services Platform through open standards [Red11]. Accessing the database layer using
methods which implement the SOA, e.g. Web services, requires the application to support
such methods. One of the main goals in this diploma thesis is to minimize the adaptations
in the DAL when migrating the data to the Cloud. Furthermore, as we can see in Figure
3.2, connection utilizing native database protocol is established directly from the business
32