Migrating e-Science Applications to the Cloud: Methodology and Evaluation - Cloud Computing with e-Science Applications - page 94

result in extensive data transfer between the OPAL services on the premises
and the database off the premises, therefore creating a potential bottleneck.
Considering this, we decided also to migrate the business logic of the applica-
tion to the cloud. Consequently, the modeling and monitoring tool was kept on
the premises, while the remaining parts of the SimTech SWfMS were moved to
an off-premises infrastructure. As a result, we not only avoid bottlenecks but
also reduce costs since for most cloud providers data transfer inside the cloud
is significantly cheaper than data transfer from and to the cloud.
The challenges we faced during this process were the following:
• which part of the system to migrate,
• what is the target system to migrate on,
• if and how to adapt the existing system to operate correctly after
the migration,
• and most important, the lack of automated support with respect to
these decisions.
To address these challenges, in this work we present a methodology that
incorporates decision and refactoring support for migration of the database
layer of applications to the cloud. For this purpose, in the following section
we focus on investigating available methodologies and decision support
systems (DSSs) for such scenarios.
5.3 Related Work
The state of the art we investigate in this section covers three aspects. First, we
review existing literature on recommendations, benefits, and use cases with
respect to the usage of cloud computing for e-science. Second, we investi-
gate available vendor-specific and vendor-independent methodologies and
guidelines for migrating either the database layer or the whole application
to the cloud. Then, we consider available recommendations and DSSs with
respect to migration to the cloud.
Mudge et al. reported an increase in speed by a factor of five on execution
times when they migrated an e-science application from the domain of geo-
physics from on premises to the cloud, considering services from Amazon
AWS and Microsoft Windows Azure (Mudge et al., 2011). Cala et al. used
cloud computing to satisfy the demand for increased computation power
and need for storing large volumes of data by migrating an existing e-science
application for predicting chemical activity to Microsoft Windows Azure
(Cala et al., 2013). The migration scenarios we are using in our methodol-
ogy not only cover enterprise use cases but also cover scientific scenarios,