Talk on Hybrid Cloud Infrastructures for Scientific Applications by Dr. Salman Toor
Date & Time: Thursday 11th December @ 2:30pm
Venue: IBM lab, SEECS
The cloud paradigm allows building a scalable computational and storage infrastructure for a variety of applications. Currently most of cloud platforms are based in a single data-center facility. This model serves well as it allows efficient resource management and security. However the model of building large computational islands still requires communication and coordination between them.
The DII-HEP private cloud, based on the OpenStack cloud suite and the Gluster file system, is designed to provide a scalable and secure private infrastructure for scientific applications. The infrastructure is evaluated with the Compact Muon Solenoid (CMS) data analysis framework, a highly challenging scientific application that requires computational, storage and network resources. For handling security issues, we have used the Host Identity Protocol (HIP). The HIP approach is portable (vendor agnostic), tenant oriented and thus can be deployed in any IaaS cloud. In DII-HEP, the HIP based security layer provides bases for building a secure multi-tenant and hybrid environment. For cloud bursting, the resources of CSC (IT Center for Science, Finland) cloud are used.
Salman Toor has done his Masters and PhD in Scientific Computing from Uppsala University, Sweden. In 2012, Toor was hired as a postdoctoral researcher in CMS program, Helsinki Institute of Physics, and IT Department, Uppsala University. His area of expertise include management, scalability and performance of distributed computational and storage infrastructures for scientific applications. He is an active member of Distributed Computing Applications (DCA) group at Uppsala University.