70 likes | 171 Views
Measurement System Spiral 2 Year-end Project Review. University of Wisconsin, Colgate University, Boston University PI : Paul Barford, Joel Sommers , Mark Crovella Staff : Mike Blodgett, Charles Thomas Students : Tiantong Yu, John Raffensperger , Ananya Das August 30, 2010.
E N D
Measurement SystemSpiral 2 Year-end Project Review University of Wisconsin, Colgate University, Boston University PI: Paul Barford, Joel Sommers, Mark Crovella Staff: Mike Blodgett, Charles ThomasStudents: Tiantong Yu, John Raffensperger, Ananya Das August 30, 2010
Project Summary • Objectives • Develop network measurement capability for GENI • ProtoGENI front end • Aggregate manager that arbitrates between ProtoGENI and sensors • Packet capture sensor systems that can run on commodity hardware • Packet transformation, authenticated/privacy preserving, and backend storage • Spiral 1 • Completed specifications • Completed test suites • Completed basic prototypes • Spiral 2 • 95% of target tasks are complete • Completed and tested major components of the system • Beginning rollout on NLR • ProtoGENI integration pending • Demoed system at GEC8 • Other • Many challenges despite professional programmers with some student involvement • Chairing GENI I & M Working Group INSERT PROJECT REVIEW DATE
Milestone & QSR Status * The GPO has not signed off on any of our spiral 1 or spiral 2 milestones INSERT PROJECT REVIEW DATE
Accomplishments 1: Advancing GENI Spiral 2 Goals • Per the GENI Spiral 2 Overview, the objectives are focused on “moving toward experimentation with specific focus on integration, interoperability and identity management. • Toward experimentation • We have developed the three key functional elements of our system (GUI, Aggregate Mgr. and Sensor) to the point where they can be deployed in the wide area. • The availability of our system enables researchers using GENI to specify and experiment, capture packets, transform the packets and archive the packets. This is unique/valuable capability. • Toward integration • Our key integration point is ProtoGENI. This has been a MAJOR difficulty over the past year. The FACT is that the many aspects of ProtoGENI and ill defined, under development or completely undefined. Eg. we are the first to try to use the ProtoGENI Reference Component Manager and are helping to develop and debug that system. • Our system has been developed with modular, well defined interfaces that can easily be extended to other, stable components of GENI. • Toward interoperability • Our system has been designed to be extensible and to interoperate with other systems • There are no active interoperability targets yet, but we are talking with UK (J. Griffioen) • Toward identity management • Our system uses ProtoGENI’s user authentication system • Link resource management is TBD INSERT PROJECT REVIEW DATE
Accomplishments 2:Other Project Accomplishments • All Wisconsin Advanced Internet Lab (WAIL) resources are available via the ProtoGENI API • Established layer 2 connectivity from WAIL to to multiple national backbones (I2 ION, NLR Framenet, ProtoGeni, Openflow), which paves the way for broad deployment going forward • Participated in all GEC’s • PI-Barford is co-chair of the GENI Instrumentation and Measurement Working Goup • Initial meeting of the GIMS Working Group had relatively light attendance, but GEC8 attendance was very strong • Active and growing specification of I & M specs. • Undergraduate involvement at Colgate University under PI-Sommers • Funds were secured from Colgate to hire under grad research assistants for the academic year and to send them to GEC8 INSERT PROJECT REVIEW DATE
Issues • The single most significant issue for our project is the fact that specifications and implementations of the ProtoGENI control framework are fluid. This is not a criticism of ProtoGENI or the Utah folks (they have a lot on their plate and are making good progress)! Fluid implementations hinder our progress toward making our capabilities available to researchers. • In order for our capabilities to be broadly available, we will need to deploy measurement sensors remotely. This is a time consuming and costly process, much of which is beyond our control. This makes timing on deliverables difficult to project. • We are spending about 2x more hours on this project than funds provide for. This implementation is very complex so managing to stated milestones will continue to be challenging. INSERT PROJECT REVIEW DATE
Plans • Remainder of Spiral 2 • Complete the ProtoGENI integration – end of Sept. ‘10 • Continue tests on sensor scalability – end of Sept ‘10 • Complete deployment on NLR – end of Sept. ’10 • Spiral 3 • MANY ideas for what is required going forward! • Continue to work with ProtoGENI to enable the I & M system to be used in that control framework • Create workflow system that enables researchers to be able to use GENI without having to be an expert in GENI • Backend tools for data analysis • Broader deployment in NLR toward the goal of supporting experiments using ProtoGENI framework • Integration with another control framework (not sure which makes the most sense) • Interoperability with another measurement system (UK group, others) • Extension of capabilities for GENI operations and management INSERT PROJECT REVIEW DATE