Table 1.
A list and an overview of the pertinent projects related to the cloud infrastructure management
| Cluster provisioning and management projects and tools | |
|---|---|
| (Sun/Oracle/Open) Grid Engine http://en.wikipedia.org/wiki/Oracle_Grid_Engine |
Enables a set of computers to be composed into a compute cluster, enabling user to submit compute jobs via a unified interface. CloudMan is currently using the most recent open source version of Sun Grid Engine. |
| Condor http://research.cs.wisc.edu/condor/ |
Enables high throughput computing on distributed compute resources, providing features such as job dispatching and monitoring. We are actively looking to integrate Condor into CloudMan. |
| Rocks http://www.rocksclusters.org/ |
Focuses on compute cluster configuration and management with no application-level integration (e.g., data sources, dependent tools, scaling) or deployment bundling and sharing. |
| OSCAR http://svn.oscar.openclustergroup.org/ |
Focuses on allowing users to install, administer, and program a dedicated compute cluster with a range of installed packages, with the default set of packages focusing on scientific computing. |
|
Cloud orchestration projects and tools | |
| Eucalyptus http://www.eucalyptus.com/ |
These are in many ways similar cloud middleware projects enabling Infrastructure-as-a-Service (IaaS) management of a compute infrastructure or a datacenter. These projects provide building blocks (e.g., virtual machines, block storage, object storage, networking) for assembling higher-level application services, as the one described in this paper. |
| OpenNebula http://opennebula.org/ | |
| OpenStack http://www.openstack.org/ | |
| StarCluster http://star.mit.edu/cluster/ |
Enables a general-purpose compute clusters to be easily deployed in the AWS environment using the command line. Although feature-full, at the moment, StarCluster operates only in the AWS context and provides no notion of deployment sharing, integration with specific applications and data sources, or a graphical management interface. |
| DeltaCloud API http://deltacloud.apache.org/ |
These provide a uniform API allowing a standardized way of programmatically communicating with a range of clouds without needing to differentiate between those. We have and are exploring options of using such libraries internally. |
| Libcloud API http://libcloud.apache.org/ | |
| CloudInitD http://www.nimbusproject.org/doc/cloudinitd/ |
Enables a contextualization hook to be made available in a given cloud instance allowing one to customize that particular instance at runtime by providing explicit system-level instructions that should be executed by the system at boot time; such functionality is an integral part of any IaaS cloud middleware. |
|
Server configuration management projects and tools | |
| Puppet http://puppetlabs.com/ |
These projects fall under the category of resource configuration management allowing one to provide a detailed recipe that is retrieved by a given machine from a predetermined server at boot time. The recipe is then executed allowing the given machine to be configured as specified in the recipe. The specified tools ensure the recipe is properly executed and a machine configured as instructed. However, once configured, these solutions do not focus on subsequent user-level application management of the infrastructure in a cohesive manner (i.e., as a compute cluster, a cooperative deployment, balancing the workload, data persistence) and thus represent a lower level of interaction with the infrastructure than what is described in our manuscript. |
| Chef http://wiki.opscode.com/display/chef/ | |
| LCFG http://www.lcfg.org/ | |
| CFEngine http://cfengine.com/ | |
| Bcfg2 http://trac.mcs.anl.gov/projects/bcfg2 | |