SaaS Clouds Supporting Biology and Medicine - Cloud Computing with e-Science Applications - page 252

TABLE 11.1
Selected Manag ement and Application Access Scripts

Shell Script Description

awsConnect
[-n number]

Connect to Amazon EC2 and start n number of Cluster Compute instants
(an EC2 cluster)

awsReady Check readiness of the EC2 cluster created for accepting instructions

awsTerminate Terminate the EC2 cluster created

awsRun

Run an HPC application in an EC2 cluster

awsTransfer Transfer file(s) to an EC2 cluster

awsCheck Check readiness of the result

awsCollect Collect result files from Amazon to a local host for postprocessing

total number of instances for the HPC cluster. The former contains all infor-
mation necessary to boot instances of software, such as operating system,
middleware, and a specific HPC application. The latter quantifies the size
of the HPC cluster, which in turn defines the maximum computing power
to be provided.
Although there is a large collection of public AMIs available from Amazon
and the EC2 community, it is the responsibility of an HPCaaS service pro-
vider to provide a customized AMI for each HPC application it supports.
In our example, all AMIs provided in the HPCaaS service library for Amazon
EC2 run the Ubuntu Linux operating system, the OpenMPI middleware,
and Linux-based open-source HPC applications. These AMIs have been
made publicly accessible. Users can access them by referencing the AMI IDs.
Consequently, the construction of an HPC cluster and the installation and
configuration of an HPC application has been abstracted into selecting an
AMI from Amazon to use.

11.4.4 Uncinus: An Application Broker and HPCaaS Cloud Solution

Uncinus integrates the HPCaaS libraries into a web platform, allowing
researchers access to Amazon EC2 resources configured for HPC. To imple-
ment the features described in the framework, a number of services are
provided. These services fall into the SaaS and HPCaaS layer previously
described (see Figure 11.4). The SaaS layer provides application broker func-
tionality through the Application to Interface Parser (AIP) and Resource
Deployment Recorder (RDR); these services write data to the mySQL data-
base management system (DBMS). The HPCaaS layer provides the function-
ality to configure cloud resources for HPC and deploy applications. Cloud
resources configured for HPC are provided by the Cloud Resource Allocation
(CRA) service, while application deployment is supported through the
Secure Data Transfer (SDT) service. The CRA service incorporates the access
scripts developed as part of the HPCaaS model (see Section 11.4.3.2) and as
such interacts with Amazon EC2 using the Amazon API.