Company Also Demonstrates Application Services Solution Integrated with Platform Computing's LSF
FREMONT, Calif., Nov. 6 - Evergrid, Inc., a provider of global resource management software for high performance computing (HPC) and next generation data centers, today announced it will demonstrate the beta version of Evergrid Load Manager at SC '07, the international conference on high performance computing, networking, storage and analysis, being held in Reno, Nevada, November 10-16, 2007.
Evergrid Load Manager targets HPC and enterprise workloads by enabling fine-grain monitoring and control of application load on multiple servers, while guaranteeing the service levels of each application. It ensures that vital applications provide timely service and can throttle down application resource usage when server capacity is needed to run higher priority applications. It also monitors application resource utilization in real time, and its reporting capability can enable improved capacity planning.
Based on Evergrid's patented operating system abstraction technology, Load Manager implements a comprehensive user-level scheduler that allows independent control of CPU, network and storage bandwidth, and memory usage of each application running on the server. By controlling these system resources at the application level, it can ensure that any application, regardless of priority, gets the resources it needs to run at any given point in time. It also provides a set of abstractions that lets data center administrators organize application processes into logical units and assign capacity to each unit.
"Evergrid's continuous availability and resource management software is designed for HPC and large data centers with expanding grids or clusters of commoditized servers. Managing to high levels of utilization in this environment, while guaranteeing customer service levels and optimized job throughput, is extremely difficult," said Fred van den Bosch, CEO of Evergrid. "Evergrid addresses both the availability and utilization issues with software solutions that optimize data center resources from the point of power-on all the way to application load management. Load Manager will be the first step in our evolution toward a fully integrated stack, with the complete range of functionality needed in the HPC and utility enterprise data center."
At SC '07, the company will also feature Evergrid Availability Services (AvS-Batch), its patent-pending high availability software, integrated with the LSF workload manager from Platform Computing. The combined solution allows massively parallelized distributed applications and single-process applications, such as Electronic Design Automation applications, to run at near 100-percent reliability in high performance computing clusters.
Evergrid AvS-Batch captures the collective state of distributed, parallel and/or long running applications, and prevents downtime due to hardware or software failures by performing checkpoint, migration and recovery of the application to ensure automatic failover across multiple nodes and tiers. AvS-Batch also ensures that when a job is pre-empted by a scheduler, the job is simply checkpointed to disc to allow a higher priority job to run to completion, and is then resumed, ensuring that it never loses a compute cycle. Stateful job pre-emption optimizes resource utilization while providing a much more deterministic completion time for applications. The integration of AvS and LSF is an example of how AvS can be integrated with leading job schedulers to provide this critical set of functionality. Evergrid AvS-Batch can be licensed separately for integration with LSF and other popular resource managers.
Evergrid will showcase Evergrid Load Manager and the integrated AsV-Batch/Platform LSF in booth #2715 at SC '07, November 10-16, 2007, Reno-Sparks Convention Center, Reno, Nev.
For more information, visit www.evergrid.com/.
About Evergrid, Inc.
Evergrid, a provider of complete server and application lifecycle management infrastructure for next generation data centers, lets massively parallelized, distributed applications run properly on high performance cluster grids, at near 100 percent reliability. Evergrid's fault tolerant application abstraction software prevents downtime, automates checkpoint, migration and recovery of applications, and scales to thousands of nodes with less than five percent performance overhead.
Evergrid's leadership team brings extensive management and technology expertise from IBM, Amdahl, VERITAS, Motorola, Tandem Computers and the Virginia Polytechnic Institute and State University. Evergrid is a private company. For more information, visit www.evergrid.com/.
Page One PR for Evergrid
Source: Evergrid, Inc.