Information Technology Reference
In-Depth Information
of them to the cloud. The e-science domain, especially the scientific workflow
community, has reported concrete benefits from utilizing cloud infrastruc-
tures for isolated use cases. In this respect, there is a clear need for a method-
ology supporting the migration of e-science applications to the cloud. There
are two key aspects that characterize e-science applications: large amounts
of data and intensive computational tasks to be performed on these data.
In this work, we focused on the former, discussing how to support the migra-
tion of the database layer of e-science applications (and beyond) to the cloud.
Supporting the migration of the database layer of an application to the
cloud involves not only considering the requirements on the appropriate
data source or service imposed by the application but also the possible need
for adapting the application to cope with incompatibilities. In the previous
sections, we presented a step-by-step methodology that considers both
aspects of the migration. To construct this methodology, we first identified
a series of functional and nonfunctional requirements from both e-science
and business domains. We then adapted the methodology discussed by
Laszewski and Nauduri (2011) to satisfy the identified requirements, result-
ing in our proposal for a seven-step end-to-end methodology for the migra-
tion of the database layer to the cloud and for the application refactoring
required as part of this process.
Then, we discussed the realization of our proposal as a publicly available
and free Cloud Data Migration Tool. The tool provides two fundamental
functionalities: decision support in selecting an appropriate data store or
service and refactoring support during the actual migration of the data. Users
of the tool can currently create migration projects, define their requirements
in terms of the migrated database layer to the cloud, describe their current
database layer, and receive recommendations, hints, and guidelines on where
and how to migrate their data. Conflict resolution is based on previously
identified cloud data patterns, and data adapters are provided, allowing for
the automatic migration of data to recommended data stores and services.
We  evaluated our proposal by migrating the SimTech SWfMS to Amazon
Web Services solutions and showed that, while useful, our methodology and
tool need further improvements.
In particular, according to our evaluation, our proposal needs to be extended
to provide explicit support for the testing phase of the migration. The Cloud
Data Migration Tool must be extended to provide sandboxing capabilities
and both functional testing for bug fixing and performance benchmarking
tools for different application workloads. These capabilities can also be used
toward supporting the optimization of the database layer after its migration.
Additional functionalities that are currently being implemented to the Cloud
Data Migration Tool, as identified in the previous sections, include address-
ing the impact of the migration to compliance, supporting more than one
source or target data stores or services and multiple migrations per project,
increasing the number of adapters available in the tool, as well as improving
the usability of the tool for scientists.
Search WWH ::




Custom Search