| Version 4 (modified by , 11 years ago) (diff) |
|---|
GENI MAX Stitching Project Status Report - Post GEC16
I. Major accomplishments
A. Milestones status
MAX: S4.d: Continue to federate the MAX Network with GENI and make it available to GENI experimenters, as well as for testing and demonstrations of network stitching in GENI
(complete)
- Implement the GENI RSpec v3
- Provide resources to experimenter credentials signed by one of the GENI slice authorities
Status:
The MAX AM has been upgraded to be compatible with GENI RSpec v3. The MAX GENI AM is also been federated with ProtoGENI and PlanetLab. Experimenters are able to utilize credentials and slices from those aggregates to instantiate experiments. We still need to update the MAX information on the Aggregate Information page on the GENI Wiki.
MAXREGIONAL: S4.e: Prototype GENI Topology and Path Computation Services
- Initial Prototype: January 2013
- Production Ready Service: May 1, 2013
- Prototype GENI Topology and Path Computation Services (previous SOW is just for a design)
Status:
We have completed an initial version of the Stitching Computation Service (SCS). This has been utilized for development of the workflow service (which is the next milestone). A description of the SCS is available here:
The SCS software is available here:
We consider this software to be beyond prototype, but not yet transitioned to operational use. This transition will happen as part of the conduct of the Stitching Testing defined here:
The stitching topology service prototype implementation is still underway. We currently have a very basic topology service prototype implemented as part of the computation service. This provides the topology information that the SCS needs, but does not currently provide a generally available query mechanism or an automated update function via GENI AM API calls.
We have also been working with the GEMINI project to utilize their topology services for stitching services. They have implemented the ability to dynamically and automatically obtain topology information from the GENI AM API as well as the existing PerfSonar Topology Service. We have not implemented a mechanism to pull the information from the GEMINI Topology service. The topology service final design will likely be either based on extending what we currently have embedded in the SCS or via utilizing the GEMINI Topology Service. During the next reporting period we will finalize the plan for the GENI Topology Service and continue with implementation.
B. Deliverables made
During this reporting period we completed the following deliverables:
GEC15 Stitching Demonstration
- geni.maxgigapop.net ====> Publications ====> GEC15
or direct
- https://geni.maxgigapop.net/twiki/bin/view/GENI/Publications#GENI_Engineering_Conference_15_O
- http://groups.geni.net/geni/wiki/GEC15Agenda/EveningDemoSession#Mid-AtlanticCrossroadsMAX
MAX Aggregate Manager support for GENI v3 RSpecs:
MAX Aggregate Manager federated with ProtoGENI and PlanetLab and available for use via user and slice credentials from these aggregates.
II Description of work performed during last quarter
A. Activities and findings
Based on the work we have completed with our prototype deployments of ION and LONI GENI Aggregate Managers, we believe we are ready for a more production level deployment. We have started working with Internet2 NOC to deploy an instance of a GENI AM that they will maintain and support on a trial basis.
We have a well defined stitching architecture which consists of the following components:
- Stitching Extensions for RSpec Schemas
- Stitching WorkFlow Function (basic and w/negotiation)
- GENI AM API Extensions for Stitching
- Stitching Topology Service
- Stitching Computation Service
Based on already completed design, implementation, testing, and demonstrations, the indications are that we will be able to implement a GENI multi-aggregate stitching capability which will satisfy the needs of the initial GENI Rack deployments as well as other sites which will have GENI Aggregate Managers. Now that the GENI AM API v3 has been finalized, we will need to update the current design for item ii (stitching w/ negotiation). These updates are primarily message type and format in nature, and do not change the planned architecture or design.
This initial capability set will likely be based on items i and iii (with basic stitching workflow). This capability will allow GENI experimenters to utilize the initially deployed infrastructure to instantiate slices which span multiple aggregates. Parallel to this initial stitching deployment, work will continue on items ii (stitching w/negotiation), and iv-v to prepare for larger scale deployments.
B. Project participants
Tom Lehman (MAX)
Xi Yang (MAX)
C. Publications (individual and organizational)
No formal publications were funded by this projects during this time period.
D. Outreach activities
The current focus of our outreach activities is thru our collaborations with other Aggregates such as CRON, ProtoGENI, ExoGENI, and GENI Rack deployment sites. We also plan to encourage more outreach to potential experimenters at University of Maryland and other universities in the MAX footprint.
We are participating in conversations with the NSF CC-NIE community to discuss GENI. We also have a NSF CC-NIE project at MAX and will be working on the integration of this with our GENI capabilities on UMD campus and MAX member institutions.
E. Collaborations
As part of the activities and milestones described here, we have had extensive collaborations with many in the GENI community. This includes CRON, ProtoGENI, ExoGENI, and GPO developers. A large focus of our milestones during this reporting period revolved around network stitching architecture, design, and software development. These activities required collaborations with other GENI aggregate manager developers.
F. Other Contributions
none.
