wiki:Gec9ClusterDAgenda

Cluster-D Meetings at GEC9, Agenda and Notes

Tuesday, Nov 2, 12noon - 4:30pm
Room: Congressional A
Meeting Chair: Ilia Baldine (RENCI)
System Engineer: Harry Mussman (GPO, Raytheon BBN Technologies)

Agenda Slides

Introductions (12noon)

Spiral 3 Goals (12:05pm, Mussman)

The following high-level goals have been established by the GPO for Spiral 3:

Experiments:

  • Use by outside researchers
  • Use involving multiple aggregates
  • Use for research experiments

Some expected experiments using Cluster D during Spiral 3:

  • Programmable Packet Networks over Dynamic Circuit Substrate (Jeff Chase, Xiaowei Yang)
  • Optical-layer experiments (BEN network, ERM sensors and actuators, IMF capabiliites)
  • Mobile wireless experiments (DOME)
  • Nowcasting experiments, using weather data from ViSE radar, processed in local cluster and/or cloud (ViSE and DICLOUD)
  • Sensor network experiments (Kansei, Neteye)
  • Robotics experiments (OKGems)

Interoperability:

  • Researchers can obtain and stitch resources from multiple aggregates
  • Researchers from other CFs (e.g., protoGENI) can obtain resources from Cluster D aggregates via GENI AM API
  • Researchers in Cluster D can obtain resources from aggregates other CFs (e.g., protoGENI)via GENI AM API

Proposed Spiral 3 Roadmap for Discussion (12:15pm, Mussman and Baldine)

Cluster D clearinghouse from Spiral 2:

  • Operational readiness
  • Connectivity by aggregates
  • Features

Cluster D connectivity from Spiral 2:

  • Control framework message flows between actors
  • Component programming via SSH
  • VLAN connectivity to Internet 2 or NLR
  • Many VLANs
  • Bridge between backbone networks
  • Ability to control setup of VLANs as part of experiment
  • Ability to control setup of VLANs as part of experiment involving multiple aggregates

Task list for Spiral 3:

  • Streamline Cluster D site setup and operation, including cluster (Eucalyptus) and connector (L2 switch, with full tag mapping)
  • Streamline connectivity to Cluster D sites on campuses, via regional networks (LEARN project)
  • Plan to allow Cluster D researchers to obtain and stitch resources from multiple Cluster D sites, connected by backbone domain, using their SM.
  • Plan to authenticate users in SM via InCommon/Shibboleth
  • With other CFs, a plan for consistent names and interfaces
  • With other CFs, a plan for resource descriptions at interoperability interfaces
  • Plan to interop with other CFs, particularly from and to protoGENI (Cluster C)
  • Plan to allow GENI (e.g., protoGENI) researchers to obtain resources via GENI AM API v1.0 (using RSpec resource description)
  • Plan to allow Cluster D researchers to include tools (e.g., GUSH) to help authorize, reserve, assign and configure resources via “ORCA SM API v1.0” XMLRPC interface to their SM (similar to protoGENI AM API, or possibly same as GENI AM API v2.0)
  • Plan for Cluster D researchers to get resources from protoGENI aggregates via GENI AM API v1.0 (or possibly protoGENI AM API) (using RSpec resource description)
  • Plan for naming and stitching L2 connectivity across Cluster C (protoGENI), and Cluster D sites
  • Plan for operations portal into Cluster D site, including status and logs, with connections to GMOC

Demos for GEC10:

  • Demo experiment where Cluster D researchers obtain and stitch resources from multiple Cluster D sites (and backbone domains) using their SM.
  • Demo experiment where protoGENI researchers obtain resources via GENI AM API v1.0, using RSpec resource description.
  • Demo operations portal into Cluster D site, for status and logs, with connections to GMOC

Demos for GEC11:

  • Demo experiment where Cluster D researchers get resources from protoGENI aggregates via GENI AM API v1.0 (or possibly protoGENI AM API) (using RSpec resource description)
  • Demo experiment where Cluster D researchers stitch L2 connectivity across Cluster C (protoGENI), and Cluster D sites
  • Demo experiment where Cluster D researchers include tools (e.g., GUSH) to help authorize, reserve and assign resources via “ORCA SM API v1.0” XMLRPC interface to their SM (similar to protoGENI AM API, or possibly same as GENI AM API v2.0)

All Figures:

Figure 1-1) and 1-2): Cluster D Config at end of Spiral 3

  • Each Cluster D site is a self-contained aggregate, which can offer resources to all of GENI using GENI AM API.
    (See figure below)

Figure 2): ORCA Site Config

  • Adding interface for Site and GMOC admin/operations
  • Adding ExpCntrlTools
  • Adding Measurement Tools (various I&M projects)
  • Adding UserWorkspaceSrvc and DigitalObjectArchiveSrvc (CNRI project)
    (See figure below)

Figure 3): Multiple ORCA Sites

  • Actors exchange ORCA messages
  • Supports stitching
  • Supports advanced reservations
    (See figure below)

Figure 4): ORCA Interop with protGENI

  • Both from and to protoGENI
  • Expect to be able to stitch both Cluster D and protoGENI resources
    (See figure below)

Figure 5): ORCA Experiment Flow

  • All steps need to be supported, per I&M architecture
    (See figure below)

ORCA Features Spiral 2 report and Spiral 3 roadmap (1pm, Baldine)

Slides

  • Upcoming Bella 2.2 status
  • Bella 2.2 Features in detail
  • Bella 3.x upcoming Spiral 3 features
    • New substrate support
    • Experiment support
    • Authentication and authorization
  • ProtoGENI interoperability
    • Connectivity planning
    • Naming and format conversions
    • Handlers and policies
    • Tools integration
  • Clearinghouse and intra-cluster connectivity
    • Status
    • Actor registry
    • Future

Break (1:40-2:00pm)

Cluster D Project Plans for Spiral 3 (2pm - 3:30pm)

Each project should prepare NO MORE THAN THREE slides for presentation, and upload to the wiki below:

1) Functions and features planned for Spiral 3:

2) Connectivity planned for Spiral 3:

  • Component programming via SSH?
  • VLAN connectivity to Internet 2? NLR? How many VLANs?
  • Ability to control setup of VLANs as part of experiment?

3) Experiments planned for Spiral 3:

  • Use by outside researchers?
  • Use involving multiple aggregates?
  • Use for research experiments?

ORCA/BEN

(see above)

DOME

(not available)

ViSE and DICloud

David Irwin for Michael Zink

Slides

KANSEI

Wenjie Zeng for Anish Arora

Slides

OKGems

(not available)

ERM

Cathy Chang, Michael Wang and Wenjia Zhang for Keren Bergman

Slides

IMF

Can Babagolu for Rudra Dutta

Slides

LEARN

Deniz Gurkan

Slides

iGENI

Joe Mambretti, Jim Chen

Sldies

Open Discussion (3:30pm)

Who is going to work with the ORCA group on the following tasks?

  • Streamline Cluster D site setup and operation, including cluster (Eucalyptus) and connector (L2 switch, with full tag mapping)
  • Streamline connectivity to Cluster D sites on campuses, via regional networks (LEARN project)
  • Plan to allow Cluster D researchers to obtain and stitch resources from multiple Cluster D sites, connected by backbone domain, using their SM.
  • Plan to authenticate users in SM via InCommon/Shibboleth
  • With other CFs, a plan for consistent names and interfaces
  • With other CFs, a plan for resource descriptions at interoperability interfaces
  • Plan to interop with other CFs, particularly from and to protoGENI (Cluster C)
  • Plan to allow GENI (e.g., protoGENI) researchers to obtain resources via GENI AM API v1.0 (using RSpec resource description)
  • Plan to allow Cluster D researchers to include tools (e.g., GUSH) to help authorize, reserve, assign and configure resources via “ORCA SM API v1.0” XMLRPC interface to their SM (similar to protoGENI AM API, or possibly same as GENI AM API v2.0)
  • Plan for Cluster D researchers to get resources from protoGENI aggregates via GENI AM API v1.0 (or possibly protoGENI AM API) (using RSpec resource description)
  • Plan for naming and stitching L2 connectivity across Cluster C (protoGENI), and Cluster D sites
  • Plan for operations portal into Cluster D site, including status and logs, with connections to GMOC

Adjourn (4:00pm)

Figures

All Figures:

Figure 1-1) and 1-2): Cluster D Config at end of Spiral 3

  • Each Cluster D site is a self-contained aggregate, which can offer resources to all of GENI using GENI AM API.



Figure 2): ORCA Site Config

  • Adding interface for Site and GMOC admin/operations
  • Adding ExpCntrlTools
  • Adding Measurement Tools (various I&M projects)
  • Adding UserWorkspaceSrvc and DigitalObjectArchiveSrvc (CNRI project)


Figure 3): Multiple ORCA Sites
Actors exchange ORCA messages
Supports stitching
Supports advanced reservations


Figure 4): ORCA Interop with protGENI

  • Both from and to protoGENI
  • Expect to be able to stitch both Cluster D and protoGENI resources


Figure 5): ORCA Experiment Flow

  • All steps need to be supported, per I&M architecture


References

Last modified 8 years ago Last modified on 11/10/10 17:51:20

Attachments (16)