UCSC's Biomolecular Department Eliminates I/O Bottleneck with Panasas
1 like160 views
The University of California, Santa Cruz (UCSC) implemented the Panasas storage system, resulting in over a 40-hour reduction in run times for their research projects. This system improved performance, allowing researchers to efficiently run complex analyses on biological data and significantly alleviating I/O bottlenecks. The scalable and high-performance nature of the Panasas solution not only streamlined operations but also enabled UCSC to maximize its Linux cluster computing capabilities.
UCSC's Biomolecular Department Eliminates I/O Bottleneck with Panasas
- 1. Customer Success Story University of California, Santa Cruz University of California, “The Panasas Storage system has reduced Santa Cruz run times by over 40 The Center for Biomolecular Sciences and Engineering at University of California, hours.” Santa Cruz (UCSC) launches interdisciplinary research and academic programs that address the scientific questions of the post-genomic era. The Center uses Robert Baertsch Research Assistant, computational, mathematical, and statistical approaches to probe and analyze UCSC biological data, from DNA to biological processes to healthcare systems. One of the Center’s major projects is the UCSC Genome Browser, a web-based tool that allows researchers to view all 23 chromosomes of the human genome from as large as a full chromosome down to an individual nucleotide. The UCSC Genome Browser integrates the work of numerous scientists in laboratories worldwide and includes work generated at UCSC in an interactive, graphical display. The Challenge tests, instead of the systems that were The UCSC Genome Browser leverages running them, was critically important. extremely fast search software that runs “Many of our programs would take SUMMARY on the KiloKluster, a second-generation years to run on a single CPU. Having a Industry: Life Sciences 1000+ node bioinformatics Linux cluster. cluster with many nodes shortens run It enables researchers to match any time to days or even hours making our THE CHALLENGE DNA sequence to the human genome research possible,” said Baertsch. “Slow Slow I/O and downtime impacted the in seconds and maps experimental data I/O and downtime can really impact run times of their Genome Browser to the reference sequence. In order to overall run times and it is critical to search tool used by scientists in their process the Browser’s huge quantity of have a storage system that can scale work to solve questions of the post- genomic era. They were searching data, the Center searched for a storage to thousands of nodes with a single for a storage solution that delivered solution that delivered high performance system image.” Finally, price is always high performance random I/O to an random I/O to a large number of cluster a major consideration for the Center. A exceptionally large number of cluster nodes. “Our KiloKluster really taxes the fundamental requirement is to deliver a nodes and one that would allow them to focus solely on their tests instead of the capabilities of a storage system,” said compelling price point. systems running them. Robert Baertsch, Research Assistant at UCSC. “To be fully effective, a storage The Solution THE SOLUTION solution in our environment needs to be The Center conducted an extended The fully integrated software/hardware able to deliver exceptional performance evaluation process including detailed solution included the Panasas® with a large number of cluster nodes.” testing with many high performance Operating Environment and the PanFS™ network-attached and direct-attached parallel file system with the Panasas DirectFLOW® protocol. UCSC’s system needed to scale in storage solutions. After thorough testing, performance as well as capacity. As a the Panasas® Storage solution was result, the Center searched for a solution selected for its ability to deliver high-speed THE RESULT that had the potential to scale as a large random I/O performance in a large cluster • Exceptional I/O Performance single pool of data. Similar to many environment, simplified management • A single namespace for simplified universities, the UCSC researchers work through a scalable, shared pool of storage cluster management on several complex projects at any one and exceptional value. The Panasas Storage • Maximized ROI from their clustered time. The ability to focus solely on their system is now connected to the KiloKluster computing environment 1-888-panasas www.panasas.com
- 2. Customer Success Story: University of California, Santa Cruz to store and retrieve reference sequences. “A typical NFS server is brought to its knees when 1000 cluster nodes are pulling data “By using the Panasas from it,” said Baertsch. “With Panasas Storage and its object- based architecture, we are able to simultaneously read from and DirectFLOW® protocol we’ve write to all cluster nodes.” been able to eliminate our I/O bottleneck.” Panasas Storage helps organizations like the Center for Biomolecular Sciences and Engineering accelerate the speed Robert Baertsch Research Assistant, and accuracy of its decisions and ultimately, lead to real world UCSC breakthroughs that improve people’s lives. Panasas Storage enables the Center to maximize the benefits of Linux cluster computing by breaking down the storage bottleneck created with The Panasas Storage system’s ease of management also legacy network storage technologies. The solution is powered by added tremendous value to the Center’s solution. The single the Panasas Operating Environment and the company’s unique unified namespace ensures administrator management object-based storage architecture. In addition to exceptional will be streamlined today and in the future. As the system performance benefits, the system enables seamless growth of capacity requirements increase, the Center is confident that a single namespace, greatly improving system manageability. the Panasas Storage system can increase in size with no Finally, by leveraging industry standard components, Panasas is impact to administrator management. able to offer this solution at an extremely competitive price. The Result The Center for Biomolecular Sciences and Engineering was able to see a significant performance boost once the Panasas Storage system was moved into production. “The consistently high I/O performance delivered by the Panasas solution enables our researchers to get their results more quickly,” said Baertsch. “By using the Panasas DirectFLOW® protocol we’ve been able to eliminate our I/O bottleneck.” In fact, for specific batches of jobs the Panasas Storage system has reduced run times by over 40 hours. Perhaps even more important, by leveraging the object-based architecture, Panasas has been able to offer the Center new ways to look at increasing overall performance. “Panasas has given us many ideas on how to scale I/O and we look forward to further experiments,” said Baertsch. About Panasas Panasas, Inc., the leader in high-performance scale-out NAS storage solutions, enables enterprise customers to rapidly solve complex computing problems, speed innovation and bring new products to market faster. All Panasas solutions leverage the patented PanFS™ storage operating system to deliver exceptional performance, scalability and manageability. PW-10-21700 | Phone: 1-888-PANASAS | www.panasas.com © 2010 Panasas Incorporated. All rights reserved. Panasas is a trademark of Panasas, Inc. in the United States and other countries.