Faster Answers
Resource allocation and utilization: Efficient resource management is a major concern for scalable systems. SGI® IRGOTM has features that allow you to achieve and maintain a high degree of productive usage across your system.
- Dynamic load balancing Allows jobs to be redirected from busy nodes to idle ones on the fly, without intervention from the system administrator.
- Weighted prioritization Allows system administrators to set different priorities for threads in accessing system resources--from real-time threads with guaranteed latencies to "weightless" threads that receive minimum priority--so systems can be run at high utilization rates with no degradation in performance to core technical jobs.
- Job limits and cpusets Allow system administrators to efficiently allocate resources to specific jobs or processes so core technical jobs can be run without interruption.
- Data locality algorithms Allow data to be intelligently placed into specific nodes for faster access times.
- Large page support Allows large amounts of data to be stored in memory to avoid bottlenecks for complex applications.
- Load Sharing Facility (LSF) from Platform Computing* Allows reliable, on-demand access to computing resources and maintains a continuously balanced workload regardless of programming model.
- XFS® and CXFSTM shared filesystems Allow filesystem scalability up to billions of gigabytes without performance degradation, with sharing of remote files at local filesystem speeds and elimination of bottlenecks inherent to copy and transfer of data.
Guaranteed real-time response: For some mission-critical applications, a late answer is a wrong answer. When timeliness is as important as accuracy, you need the real-time features of IRGO.
- REACTTM real-time extensions* Provide consistent guaranteed latency and response times.
- Guaranteed-rate I/O (GRIO) Provides consistent guaranteed I/O throughput rates for real-time functionality in I/O bandwidth.
Productive uptime: Failures sometimes happen, but with IRGO they aren't catastrophes. These IRGO features help shorten your time to solution by maximizing the productive usefulness of your system.
- Checkpoint restart (CPR) saves an application's work in progress as it runs. If a running application is stopped for any reason, it can be restarted from its most recently executed "checkpoint," rather than from the beginning of the run, so all compute cycles are productive ones.
- Embedded Support Partner (ESP)* automatically notifies SGI support personnel of system outages with remote diagnostic information, so SGI can get your system running again as quickly as possible, sometimes before you are even aware of the outage yourself.
* Purchased separately
