Efficient Web Search Performance and Storage Capacity Scaling Using NFS-RDMA and InfiniBand
Dr. Ekechi Nwokah, Storage Architect, Alexa Internet

Web search and data mining applications developed by Alexa Internet, a subsidiary of Amazon.com, demand multi-terabytes of storage capacity and technology that enables cost-effective scaling.  This case study explains how InfiniBand has proven to be an excellent storage interconnect for Alexa, enabling them to grow their storage and compute resources in an independent and non-disruptive way, while also ensuring highest resource utilization and performance. 

As a result of participating in this session, attendees will be able to: 

• Describe the business and technical challenges related to crawling, storing, and analyzing the web. 
• Outline the reasons for why Alexa chose InfiniBand as a storage and compute clustering interconnect technology
• Determine how use of InfiniBand-based NFS over RDMA helped Alexa deliver the storage performance and scalability requirements for their web search and data mining applications
• Assess how InfiniBand-based storage and clustering helped Alexa to achieve optimized asset utilization while enabling independent and non-disruptive capacity and performance growth, irrespective of workload