High Performance Wireless and Research Network

Research Overview

Research Lead: Dr. Hans Werner Braun, Principal Investigator, HPWREN, San Diego Supercomputing Center and Greg Hidley, GreenLight Project Manager, Calit2.

The HPWREN and GreenLight teams developed a strategy to improve access to HPWREN image data. The GreenLight team put together an infrastructure upgrade implementation plan for this strategy designed to improve performance, control, and reliability of access to HPWREN data as well as provide increased infrastructure reliability and data redundancy. The hardware procured, provided and setup includes Sun SunFire 4540 high performance storage servers (Thumpers), an A10 Load Balancer, and Cisco Cache Engine. The upgraded infrastructure also includes multiple data paths between HPWREN data and web servers.

We copy image content from HPWREN cameras to a Calit2 Sun Thumper, and then replicate it further to three other GreenLight Thumpers. All Thumpers are on the OptiPuter network, but distributed across campus (two are in Calit2 1201 and 2 are in the Project GreenLight SunMD adjacent to the Pharmacy building). Web servers have been setup on the Thumpers and the A10 load balancer has been installed to distribute web requests across multiple servers.  In addition to pulling content from multiple servers (increasing speed and availability), the load balancer and WAE engine optimize TCP traffic as well as provide caching, and data compression offload to improve performance. The load balancer can also prioritize user requests via IP addresses, giving access priority and predictable response to emergency responders.

To summarize, in Year 3 Project GreenLight proposed and implemented the following enhancements to the prototype upgraded HPWREN web service using GreenLight hardware and services in the Calit2 building and the SunMD:

  •  Data availability improvements by prioritizing and distributing the service loads among multiple Calit2 web servers and network path
  • Web performance (data access) improvements by using multiple resource servers, high performance RAM caching, compression offload, TCP optimizations, and network path
  • Data security and access control improvements using IP based access prioritization (by controlling connection paths and transfer rates across managed communities)

We anticipate public usage beginning in April 2011, followed by the usage of GreenLight monitoring tools to determine achievable energy efficiencies using the new infrastructure.