Grid Computing

What is Grid Computing?

Grid Computing is a parallel computing architecture that breaks down and distributes a single task among multiple nodes for increased throughput. Our framework specifically applies the resources of many independent nodes to solve large-scale document processing requests over a high bandwidth local network. Tasks involving large collections of electronically stored information (ESI) are broken up into as many small and independent processes as possible and sent to available nodes for maximum throughput. The most common and logical application of this is TIFF conversion and Optical Character Recognition (OCR), but grid computing also has its applications in tasks such as native file processing and embedded object extraction.

How does it work?

First, the Ignited Grid Application submits a job directly to a specified ‘Grid Manager’. The Grid Manager then splits apart the project and queues it as multiple tasks for completion. Each task is then assigned to an individual ‘Executor’ to be processed local to its machine. When a task is finished, the executors report their ‘available’ status and are immediately assigned the next task, until all sections have been completed. If jobs running on another manager finish and additional executors are available as a result, they are simply assigned to the next manager and begin receiving any associated jobs. Due to the scalability of the system, an infinite number of jobs can be completed -- allowing us to handle jobs of 10GB or 10TB in the same manner.

Advantages of Grid Computing Reduce human error These documents are extremely important to your case. The risk of error is minimized by limiting the amount of human intervention. Increased throughput Our servers can collectively process hundreds of gigabytes of electronic files in a single day. Spend more of your valuable time working and less time waiting. Efficiency creates savings With our high level of automation, our system runs lean. Not only does this mean greater throughput, but also more savings for you and your clients. No such thing as down time. Our grid processing system has been engineered from the beginning to eliminate any potential down time. By utilizing dual management nodes with load balancing and automatic fail-over, each with redundant data storage and power supplies, we are able to resolve hardware issues without interrupting any active jobs. Also, since our system uses countless nodes for processing, there is never a moment when “the system is down”. If one or even two computers fail, the overall throughput is only marginally impacted. In addition, if a processing node failure occurs, we simply slide it out of our rack and slide in a new node on the fly, returning the system to full capacity in a matter of minutes, while not disturbing any running projects.