DRAFT
The Boston Cloud Declaration
Purpose
With the advent of cloud computing capabilities in support of scientific research, the global research community is slowly but surely migrating research workloads from locally managed (on premise, private cloud) solutions to ones supported in shared public or open-cloud environments.
For scientist in the OpenStack community and the larger “OpenCompute” community, moving away from locally managed environments can be a challenge – particularly given the sometimes idiosyncratic development of OpenStack in addressing local researcher needs. With the greater ease of access to data globally, researchers are reaching well beyond their institutional or geographic boundaries – leveraging improving transport capabilities to move data to local systems for analysis and data-mining.
Almost simultaneously, the rapid growth in the volume of data and its globally distributed nature across multiple research communities raises the question of why move the data – if the problem is simply that of code differences between and among research clusters.
Finally, the emerging high-performance compute capabilities of meta-cloud providers have cracked open the door to greater access to scientific computing resources for a broader range of academic and non-academic research and scholarship communities. Historically, research compute clusters have been manned by technologically savvy scientists who self-selected or were volunteered into a role of support providers for the remainder of the university’s research faculty. In the past, the benefit of being at a major research institution was access to the right level of resources and technical support. While the ability for meta-cloud providers to support high-performance/high-throughput cloud computing is a positive sign for the democratization of research capability, it does open an area of new challenge, namely, how will such technology services be supported across the now MUCH larger research community.
As part of the kick-off of the OpenStack Scientific Working Group these sort of question led to a discussion about the need for federated scientific research clouds and the development of a community of practice that would include both academic and “commercial” providers of scientific research technology infrastructure. It was felt that the ambition of balancing innovation against efficiency and accessibility would be best served by bringing all of the various perspectives together.
Building on the Berlin declaration for Open Access and precedent from the “Open Space Technology” approach, the OpenStack Foundation and leadership from the OS Scientific Working Group have identified the need to hold a more in-depth discussion of the challenges and opportunities of realizing a globally scoped federated scientific research cloud. To that end, and under the working title of “The Boston Cloud Declaration,” the group has targeted the 2017 OpenStack Summit in Boston to hold a two-day workshop to discuss the issues, reach consensus on how to move forward and begin actively working towards the establishment of a federated scientific research cloud.
Given the nature and scale of the project, it is expected that in addition to the scientific research computing community, the active participation of the national research and education network community will essential to moving forward. Building on efforts such as those by Internet2 in establishing the InCommon Federation (Shibboleth) and Eduroam as “community standards of practice,” the ambition is to leverage support from NRENs for global access to scientific data (perhaps a Global Scientific Data Fabric), facilitation of the inter-operability of open compute for use against scientific workloads and federating identity management globally (already underway through Internet2’s TIER efforts).
Tenative Dates and Venue:
MIT (specific venue TBD)
May 15th-16th, 2017 (alternative? – say May 13-14?) – idea is to have it either before or after the Boston OpenStack Summit (May 8th-12th 2017)
Tenative Agenda:
Day 1:
09.00-09.30: Arrival and Coffee
09:30-10:30: Welcome and Participant Introductions.
10.30-11.00: Orientation (Agenda and Ambitions)
11.00-Noon: Keynotes – Outlining the need and scope of issues (Speakers TBD)
Noon to 13.00: Lunch
13.00 to 14.30: Moderated Group Discussion
(Objective: Consensus on areas of focus and participating constituencies)
14.30 to 15.30: Topical Session 1 – Scope of Federation (focused discussion/deep dive)
15.30 to 15.45: Break
15.45 to 16.45: Topical Session 2 – Role of NRENs – Network, IdM (focused discussion/deep dive)
16.45 to 17.30: Summary and Initial draft of declaration (and identification of next steps)
BREAK for general participants until dinner at 19.30
19.30: Dinner
Day 2:
08.00-09.00: Arrival and Coffee
09.00-09.30: Moderators summary from Day 1
09.30-10.30: Topical Session 3 – TBD (focused discussion/deep dive)
10.30-10.45: Break
10.45-Noon: Drafting and Consensus on Cloud Declarations
Noon-13.00: Agreement on Next Steps (leadership/participation roles/commitments).
13.00-13.30: Concluding remarks.
13.30: Post-meeting Luncheon
Follow-up might include: