Startup Evergrid has announced a virtualisation platform designed to provide application reliability on clusters of Linux servers.
The company said its Availability Services for application recovery and restart and Resource Manager for workload scheduling are the key components of Evergrid Cluster Availability Management Suite (CAMS).
The CAMS platform is designed to be either a standalone replacement for resource management software in the corporate data centre or for integration with existing resource management platforms such as Platform Computing LSF and Altair's PBS Pro.
The software inserts an application virtualisation layer between the operating system and the application, without requiring any modification to the application. CAMS runs on physical servers and as a guest on virtual machines created with software from VMware and Xen.
The checkpoint/restore feature of the Availability Services lets users "checkpoint" a job running on the resource manager so it can be restored in case of a failure. In addition, jobs can be stopped if the cluster is needed to run a second job and then restarted when that job is completed.
"Our goal is to manage all servers like they were one entity," says Dave Anderson, CEO of Evergrid. Anderson says the company is targeting the software, which can scale to tens of thousands of nodes, at computing intensive vertical industries such as manufacturing, financial services, pharmaceutical and petrochemical research. The company is developing a version for High Performance Enterprise Computing (HPEC) and online transaction processing (OLTP).
CAMS is priced per node with a 100 node installation priced at under $200,000.
Analysts say Evergrid is targeting an emerging need for organisations that are putting more critical applications on commodity servers running Linux.