Proposed instances

The type of instance you may have access to depends directly on the project (as defined by Openstack). It is set during the cloud access request. There are three types of instances optimized for specific use cases:

HA type

This type of instance is used for teh implementation of production services. It responds to a need for high availability.

Impact on the implementation:

  • the hypervisor hardware is highly available with maximum component redundancy;
  • the instance storage is centralized, organized by distributed file systems and hosted on equipment optimized for highly available storage (N+2 on disks and N+1 on servers);
  • paths between the hypervisors and the rest of the infrastructure are redundant on ACTIVE/ACTIVE 10Gbps (VM network and storage network for VMs);
  • instances may be migrated at once to perform hypervisor maintenance without service interruption;
  • instances may be restarted on another hypervisor immediately when a hypervisor is down.

Computing type

This instance type is used for HTC workloads. It excludes interactive uses and services.

Impact on the implementation:

  • hypervisors are optimized for CPU performance and are typically configured as computing platform worker node;
  • hardware redundancy is minimal, loss of a VM should never be a problem for the user;
  • storage is local to limit the impact of an intensive workload on other instances of the same cluster. Also, to the extent that the authorities do not require a high availability, the mechanisms of migration are useless.

R&D type

This type of instance is used for research and development.

Impact on the implementation:

  • the availability is halfway between the two previous types and thus excludes the setting up of highly available services on the instances;
  • storage is local.