The Carina Environment Manager is a component to handle the deployment of interconnected multi-VM application services on top of the OpenNebula IaaS platform. It supports the automated creation and run-time scaling of multi-VM application environments according to policies. It leverages the OpenNebula contextualization framework to setup clusters of VMs in a master-slave configuration or a set of workers with an IP load-balancer in front. Policies can be defined to control how VMs are added or removed based on manual, time of day, or application load-based triggers.
The system will ensure that an appropriate number of VMs are started even in the presence of hypervisor or data center failures to meet application service requirements. Environments can be aggregated so that multi-tier services consisting of web, app, caching, database clusters can be interconnected and deployed as unit. A variety of sample integrations with distributed middleware and tools like Tomcat, CloudFoundry, Jenkins, HBase are provided.
For service developers, an IaaS platform and layered tools like this can help turn their applications from static, manually configured systems that take weeks or months to deploy and change into dynamic and adaptive systems. Infrastructure becomes just another set of APIs that you use in developing your service so that the infrastructure will scale as your service workload evolves.
- Automated deployement of clustered services (load-balanced worker pool or master-slave deployment pattern)
- Automated deployment of multiple inter-dependent clusters
- Integration with Balance software load-balancer with open hooks to support other load-balancers
- Auto-recreate VMs that disappear, fail or go into unknown state due to hypervisior or transient network/storage issues
- Auto-scaling of environments based on time of day or day of week
- Auto-scaling of environments based on load (cpu by default with support for custom metric plugins)
- Prioritization of services and resource pools into 'gold','silver','bronze' classes.
- Preemption and reclaim of resources from lower classes when higher priority classes required them
- Active-Active and Active-Passive DR support where environments in surviving data centers are scaled up in response to datacenter outages
- Centralized management and failover of the oneenv controller
Usage: oneenv [options]
-c, --create STRING Create environment from configuration -r, --remove STRING Remove environment and delete VMs -u, --scaleup STRING Scale up environment by adding VM -d, --scaledown STRING Scale down environment by removing VM -U, --upload STRING Upload config.rb, vm template or plugin to server -D, --download STRING Download log file from server -J, --jobs List all jobs -v, --vms List all VMs -k, --envvms STRING List all VMs for environment -S, --services List all services managed by Global Scheduler -R, --requests List all requests in Global Scheduler -P, --pools List all pools in Global Scheduler -j, --json Display verbose JSON output -o, --configs List all environment configs -e, --envs List all environments