Managing Data Intensive Challenges with a Science DMZ

The network of a campus or a research enterprise supports multiple business operations: Network traffic associated with administration, financial, human resources, email, etc. The network must be operated with security that protects the financial and personnel data of the organization. At the same time, these networks also support scientific research processes. Researchers depend on the network infrastructure to share, store, and analyze research data from many different external sources.

In general, campus networks are not engineered to support the data movement requirements of data intensive science. Data intensive science applications experience poor performance on general-purpose networks. Campuses serve multiple communities of interest. As a result, they are faced with the following challenge: How can campus networks be adapted to optimize flows from science applications without impacting the operation of their production networks?

This talk explores the concept of the Science DMZ as a methodology to reengineer the campus network to support the needs of multiple communities of interest, in particular, the science research community conducting data intensive science, alongside production operation. The implementation of a Science DMZ at FIU’s campus is described. It is a component of a project called FlowSurge: a coordinated and coherent approach to bridging among different cyberinfrastructures by building a Science DMZ at FIU that can be tailored to support the needs of high-performance science applications and data-intensive flows.

Researchers, from their laboratory, will be able to (1) generate science flows to shared CI or CI- connected resources on the campus; at other campuses; or at regional, national or international sites. (2) Orchestrate and participate in “at-scale” experiments by using FlowSurge to tailor the end-to-end path for optimized high-performance science applications and data-intensive flows. FlowSurge will enable IT and network staff at FIU to partition the campus network infrastructure to support Science flows without causing impact to general-purpose production flows on the campus.

See slides ...

FaLang translation system by Faboba