wiki:GeniArchTeam/11-3-2011

GENI Architecture Team meeting

Date: November 3, 2011

Attendees: Rob Ricci, Marshall Brinn, Chip Elliott, Nick Bastin, Larry Peterson, Henry Yeh, Bryan Lyles, Tom Mitchell

Summary:

The first meeting of the GENI architecture group was held over breakfast at GEC12. The charter of the architecture group had been previously vetted by the members over email and approved. The need for a concise definition of GENI and its purpose was suggested. Details of how the group would meet, communicate, identify and nail down issues were discussed. The intent of the group was clarified as the group that resolves open issues on the GENI architecture and assures the the ongoing developing GENI system conforms to the architectural plan developed by the group. The first priorities of the group, represented in action items, were to clarify Spiral 4 priorities for the architecture and development effort.

Details:

Chip opened the meeting by saying it is time to nail down the architecture. GENI needs working and interoperable software. The architecture team will define and document the architecture, and ensure that the software complies with the architecture. A lot of the work will happen between GECs and there will be a formal meeting (with published minutes) at each GEC.

There was only brief discussion about the charter per se - it was accepted by acclamation. [Note: the charter is attached to this page.]

Marshall pointed out that the architecture team needs to be the "someone" in the sentence "someone needs to decide this." The architecture team also needs to be looking longer term - 2 GECs or 5 GECs in the future. Many in GENI only look to the next GEC, but the architecture team needs to develop a longer term road map. Marshall also recommended that the GENI documents need to be updated and/or archived. Many are old and possibly out of date. They reflect hopes and plans that are either reality, still planned, or no longer planned.

Marshall's intuition is that we need a system; an instantiation of the architecture. Bryan added that we should identify holes in our current work, in essence defining opportunities for future work.

Marshall suggested defining two elevator speeches, one outward facing and one inward facing.

On the topic of receiving inputs from the GENI community, Nick suggested an RFC-like process so that the architecture group isn't writing all the documents, and so that there is a process in place for inputs.

We briefly discussed establishing a private email list with private (ie. accessible only to members) archives. Public communication with the architecture group should occur on "dev@geni.net".

We discussed having standing face-to-face meetings at each GEC and regular teleconferences in between, probably bi-weekly. [Note: A Doodle invitation for the next meeting was sent out 11-18 for a meeting right after Thanksgiving, week of 11/28-12/2].

Additional Comments:

Bryan Lyles commented that in addition to interoperable software, we need a definition of GENI that can be shown to other people. For instance, a simple set of slides - the components, their relationships and their functions. GENI should make it clear to the larger community what the GENI community has created.

Bryan would like GENI to have a standard elevator speech. His on-the-spot draft of one was, "GENI is a system for building custom networks; custom combinations of networks& computation. One use of GENI is network experimentation."

Action Items:

  1. Tom took an action to set up the mailing list of architects. [Status: DONE. The list is architects@geni.net. ]
  1. Marshall took an action to start defining Spiral 4 architectural themes and goals (Including challenge problems). [Status: BEING REVIEWED. Those were published to the architects group with the 11-18 Doodle invitation and are attached to this page. ]
  1. Marshall took an action for drafting a high level description of GENI: "What is GENI and what is its value proposition". [Status: BEING REVIEWED. A draft has been completed and is being reviewed by Henry and Chip before release to NSF and Architects. ]
  1. Marshall took an action to set up next meeting for late November. [Status: PENDING. A Doodle invitation for the next meeting was sent out for the week of 11/28-12/2, still waiting on some responses to nail down date. ]
  1. Marshall took an action to propose topics for next meeting. [Status: DONE. In the invitation, two subjects were discussed. 1. The design of the Clearinghouse and its role in the "Bad Actor" challenge problem and 2. Changes to the RSpec (advertisement and possibly request) to allow for discovery of assets by attribute/requirement and not by explicit URN]

Next Steps

  1. Review draft spiral 4 architecture themes
  1. Establish next meeting date and agenda.
  1. Prepare for discussion on proposed meeting topics (i.e. Clearinghouse design and RSpec attribute changes).
Last modified 12 years ago Last modified on 11/22/11 17:19:31

Attachments (2)

Download all attachments as: .zip