| |
| | Penguin Computing - Linux Servers & Clusters - Scyld ClusterWare HPC |
 | | The operating environment deployed to the compute nodes is provisioned “stateless”, directly to memory. |
 | | Because the compute nodes run a lightweight, in-memory distribution, they can be provisioned rapidly so users can flexibly add or delete nodes on demand, in seconds, making the cluster extraordinarily scalable and resilient. |
 | | Further, as depicted in the architectural diagram above, it is possible to configure Multiple Masters, each with their own default set of compute nodes, such that the Masters coordinate in the sharing of compute node resources on demand, as in a “cluster of clusters” and can provide failover backup to each other. |
| www.scyld.com (883 words) |
|