Changes between Initial Version and Version 1 of LAMP-FirstQuarter2010

04/30/10 10:45:06 (12 years ago)
Vic Thomas



  • LAMP-FirstQuarter2010

    v1 v1  
     3= LAMP Project Status Report =
     5Period: Jan 2010 - Mar 2010
     7== I. Major accomplishments ==
     9=== A. Milestones achieved ===
     11=== B. Deliverables made ===
     14== II. Description of work performed during last quarter ==
     16=== A. Activities and findings ===
     18'''Activity:''' Ongoing discussions with Utah’s ProtoGENI team regarding topology representation in ProtoGENI’s RSpec schema.  [[BR]]
     19'''Description:''' LAMP project members believe that leveraging the existing topology schema (referred to as UNIS schema in this document) used by projects like perfSONAR, Inter-Domain Controller Protocol (and thus Internet2’s ION, OSCARS and GÉANT2 AutoBAHN), Phoebus, among others, to represent the topology elements in GENI’s RSpec and related schemas would provide several benefits (e.g. a more general and tested topology model, easier integration with Measurement and O&M planes). [[BR]]
     21 * Utah’s ProtoGENI team largely agrees on the benefits of such a step, but wants to iron out all the details of how the schema will look like and operate and of the efforts for transitioning from the current RSpec v2.
     22 * A tentative RSpec schema implemented as an extension of the UNIS schema was created for GEC7 and some of the example messages from ProtoGENI were translated. Rob Ricci liked the way it looked in general, further discussions are needed.
     23 * Consensus was reached in the group to provide an alternate extension mechanism on the UNIS schema to reduce the disruptive impact of transitioning from RSpec v2 to a new UNIS-based schema, and to allow for the desired properties of ProtoGENI RSpec extensions defined in [RSpecExt]. The general mechanism has been defined and work is on-going to finish the resulting schema and examples to present to the acceptance of the larger community.   
     25'''Activity:''' Network monitoring of the ProtoGENI backbone using perfSONAR. [[BR]]
     26'''Description:''' We have started a measurement infrastructure of the ProtoGENI backbone using perfSONAR. The infrastructure is currently retrieving SNMP data from the HP switches in Internet2’s Washington, Salt Lake and Kansas !PoPs (see [Integration]). The data is made available to Utah’s ProtoGENI team through perfSONAR’s SNMP MA and related GUIs.  [[BR]]
     27'''Findings: '''
     28 * Rob Ricci from Utah’s ProtoGENI team liked the system, specially the interactive graphs.
     29 * There’s currently almost no data flowing through ProtoGENI’s backbone links (i.e. almost no interaction between ProtoGENI sites).
     31'''Activity:''' Collaborating with other I&M projects on a common GENI I&M architecture and measurement and topology schema. [[BR]]
     32'''Description:''' One of the overarching goals of the LAMP project is to leverage perfSONAR to create interoperability between the different I&M systems, and work towards a unified measurement architecture for GENI. [[BR]]
     33'''Findings: ''' [[BR]]
     34 * GENI’s I&M Working Group is focused on the (same) objective of identifying common components in I&M systems proposed and creating a GENI I&M architecture. The need for a common and extensible schema for measurement data storage and exchange has been identified.
     35 * A presentation was made by LAMP project members on GEC7 regarding the different aspects of perfSONAR that are applicable to GENI’s I&M architecture.
     36 * Ongoing discussions with Harry Mussman have identified great interest of leveraging perfSONAR to the initial design of GENI’s I&M architecture, especially for the measurement schemas.
     37 * During GEC7 we talked individually with some of the I&M projects of Cluster C regarding integrating their systems with perfSONAR protocols and schemas. The schema was generally a consensus and further involvement will take place to help these systems transition (main collaboration will be with the InstTools and OnTimeMeasurement projects). Architectural components and protocols generated some concerns regarding overall applicability (e.g. is the granularity of representation in perfSONAR adequate to GENI) and the effort necessary to adapt.
     38 * Architectural components present on the draft for the general I&M architecture that do not currently exist in the perfSONAR framework have been identified as useful and potential extensions to the framework. Further discussions with the perfSONAR community are needed.   
     40'''Activity:''' Development of a perfSONAR-based I&M system for GENI experimenters. [[BR]]
     41'''Description:''' A proof-of-concept I&M system will be developed using the current perfSONAR framework (architecture and services) to show how perfSONAR adapts to the GENI environment. This system will provide great flexibility to experimenters in choosing performance metrics ranging from SNMP data to active measurements to mix-and-match based on the experimenters’ needs.    [[BR]]
     42'''Findings: '''
     43 * Identification of new GENI specific components (which might be absorbed by the general perfSONAR community eventually) have been identified within a draft of the I&M systems architecture.
     44 * A cluster has been allocated at the University of Delaware to become a ProtoGENI site and work as a testbed of the I&M system. Installation of the Emulab/ProtoGENI software is ongoing and is expected to finish soon.
     45 * Research was done regarding setting up a “portable”, self-contained development environment to increase the speed of development cycles. We are currently leaning towards using Open vSwitch (OVS) with VirtualBox (for notebooks) or XenServer virtual machines to emulate a ProtoGENI site that can reside in a single machine. There is still ongoing work to verify the feasibility and amount of effort needed for this solution (e.g. Emulab will need to be adapted to support OVS).
     47=== B. Project participants ===
     48The following are active participants in the project:
     49 * Martin Swany – PI
     50 * Eric Boyd – Co-PI
     51 * Guilherme Fernandes – Research Assistant
     52 * Ezra Kissel – Research Assistant
     54=== C. Publications (individual and organizational) ===
     55No publications done so far.
     57=== D. Outreach activities ===
     58No outreach activities.
     60=== E. Collaborations ===
     61Collaborations are ongoing with Utah’s ProtoGENI Team, the Instrumentation Tools project from the University of Kentucky, and GENI’s I&M Working Group. Further collaborations are expected with other I&M projects of Cluster C.
     63=== F. Other Contributions ===