Figure 3

Modular architecture of CloudMan. The architecture of CloudMan is based on separation and subsequent coordination of otherwise independent components: the machine image, a persistent data repository, and persistent storage resources (i.e., snapshots). The machine image is characterized by simplicity; it consists only of the basic services required to initiate the application unit deployment process. The persistent data repository lives independent of the machine image and is used to provide instance contextualization details, such as, boot time scripts that define which services should be started. Lastly, persistent storage resources or snapshots are used as the storage medium for tools, libraries, or datasets required by the tools. Once instantiated, those components are aggregated by CloudMan into a cohesive operational unit. Because they are not modified during the life of a cluster, the persistent storage resources are deleted upon cluster termination. All of the cluster settings and user data are preserved in user’s account and will be reused on next cluster instantiation. The Galaxy CloudMan developers maintain items in blue; items in green are (currently) maintained by the CloudBioLinux community (http://www.cloudbiolinux.com/), while items in gray are private to a user.