Forgot the attachment, of course. -- Rob
On Jan 11, 2018, at 10:32 AM, Ross, Robert B. <rross(a)mcs.anl.gov<mailto:[email protected]>> wrote:
Hi all,
I've chatted with all of you previously about this, but I've been lax in getting moving. Lucy has asked me to organize a workshop to help define priority research directions for scalable storage and I/O (SSIO), ideally leading to a call for proposals some time in 2018. Additionally, there is some interest from Almadena on the NSF side to restart an inter-agency activity in this space. We will be the organizers that drive the ASCR process forward. We'll see also how things develop on the inter-agency front and what role we might play in that process.
Our goal is to organize our thoughts about future research directions, produce an initial document capturing that thinking and describing state of the art, plan and hold a workshop to identify a set of priority research directions, and produce a final workshop report capturing this information.
My initial plan is based off the experiences from the old HEC-FSIO process, the more recent 2014 ASCR SSIO workshop (for which there is a 2015 report), and the most recent Extreme Heterogeneity workshop planning (for which we will have the workshop soon). All of this is up for discussion/negotiation.
Summary and action items.
- Think about what you think the big issues are in SSIO research for enabling DOE science, talk with your team and colleagues.
- Start putting together the "virtual" summit materials.
- Look for the Doodle poll for a summit date.
- Look for a Doodle poll for a time we could regularly meet to discuss progress.
Details follow. Thanks in advance for your help in making this a success!
Rob
---
Organizers. Here's our team:
- Gary Grider (LANL)
- Scott Klasky (ORNL)
- Glenn Lockwood (LBNL)
- Kathryn Mohror (LLNL)
- Lee Ward (SNL)
Lucy Nowell is our ASCR contact, and I expect that Laura Biven will be included and possibly take over as contact before we are done.
We have a mailing list: ssio-2018(a)lists.mcs.anl.gov<mailto:[email protected]> for discussion. You have been added to the list. Please let me know ASAP if for some reason you cannot help.
Topical areas and initial discussion. Five focus areas were discussed for further investigation at the prior workshop.
- HW/SW architectures
- Metadata, name spaces, and provenance
- Supporting science data
- Integrating with external services
- Understanding SSIO systems
I propose that we do a "virtual" summit (i.e., videoconference) for 3/4th of a day, rather than flying somewhere for a day to meet in person. Each organizer would present their view on challenges as relate to mission and discuss if/how they think we should adjust the topical areas in response to changes since 2015. I think of these as 20 minute presentations that are devoid of our own research specifics/propaganda and focus on these big picture items. From this we would select a set of topical areas used to organize the factual status document (see below).
Factual status document. We are modeling our process after the BES "basic research needs" approach (see attachment), with adaptations to the scope of the area and ASCR needs. This process includes the generation of a factual status document (FSD) prior to the workshop. There is a separate group working down this path for an Extreme Heterogeneity FSD and workshop, too. The FSD can be thought of as the background sections of the final workshop document, and that's how we will approach it.
With this in mind, we will organize the document around our initial guesses at topical areas, understanding that the topical areas may have to be adjusted in the final report as a result of discussion. The goal would be to complete the following for the FSD:
- Introduction
- Science Challenges
- Technology Challenges
- Knowledge Gaps -- what we need to know that we don't know
- Research Needs
- Topic 1
- Definition/summary of area
- Current status
- Topic 2 ...
- Appendix: References
This is likely to be a 50 page document.
This document would be partitioned into sections to be managed by individual organizers. Rather than requesting white papers from potential attendees, organizers would invite experts in respective areas to participate in the development of requisite sections. It might be easiest to develop this document using an online system such as Google Docs, understanding that eventually it would be turned into a Word document and finally into a PDF for publication as the workshop report.
Position papers. In the Extreme Heterogeneity planning, we asked for position papers. We got over 100 of them. Parsing them has been a real challenge for the organizers. We should think about what approach we might employ for gathering input from the community prior to the workshop and what we would do with that material prior to the workshop vs. incorporating into the final report.
Workshop. We will shoot for around 60 attendees, maybe 30/20/10 lab/university/industry. This helps keep the justification and planning fairly simple. We would plan for a 1.75-day workshop to facilitate covering five topical areas as separate, single-track discussions. If for some reason we end up needing only 1.5 days for the meeting proper, this would leave more time for the organizers to meet and discuss report completion.
We will (mostly) assume that attendees have read the FSD prior to arrival at the meeting. Never the less, we will spend 1/2 day bringing the attendees up to speed, focusing on the background material from the FSD. This is similar to the opening plenary session portion of the BES workshop structure.
The remainder of the meeting could follow the model from the 2014 SSIO meeting: 90-minute moderated discussion sessions, with notes captured live in PPT. With 1.75 days we can accommodate 5 total discussion topics. We will place a topic of broad interest last to encourage attendees to participate for the full meeting. These sessions will identify current status in the area, its CS challenges, its opportunities, and its research needs, reflecting the breakout session portion of the BES workshop structure. Because the meeting is single track, there is no need for outbrief sessions (i.e., the participants will have been present for all discussions).
Report. The final report will be generated by augmenting the FSD with material summarizing the outcomes of the workshop. We will target a 75-100 page document (the 2015 report was approximately 90 pages). The outline for the report will look something like this:
- Executive Summary
- Introduction
- Science Challenges
- Technology Challenges
- Knowledge Gaps (need to understand what this typically includes)
- Research Needs
- Topic 1
- Definition/summary of area
- Current status
- CS challenges and opportunities
- Priority research directions (3-5)
- Topic 2 ...
- Priority research directions
_______________________________________________
ssio-2018 mailing list
ssio-2018(a)lists.mcs.anl.gov<mailto:[email protected]>
https://lists.mcs.anl.gov/mailman/listinfo/ssio-2018
Hi all,
I've chatted with all of you previously about this, but I've been lax in getting moving. Lucy has asked me to organize a workshop to help define priority research directions for scalable storage and I/O (SSIO), ideally leading to a call for proposals some time in 2018. Additionally, there is some interest from Almadena on the NSF side to restart an inter-agency activity in this space. We will be the organizers that drive the ASCR process forward. We'll see also how things develop on the inter-agency front and what role we might play in that process.
Our goal is to organize our thoughts about future research directions, produce an initial document capturing that thinking and describing state of the art, plan and hold a workshop to identify a set of priority research directions, and produce a final workshop report capturing this information.
My initial plan is based off the experiences from the old HEC-FSIO process, the more recent 2014 ASCR SSIO workshop (for which there is a 2015 report), and the most recent Extreme Heterogeneity workshop planning (for which we will have the workshop soon). All of this is up for discussion/negotiation.
Summary and action items.
- Think about what you think the big issues are in SSIO research for enabling DOE science, talk with your team and colleagues.
- Start putting together the "virtual" summit materials.
- Look for the Doodle poll for a summit date.
- Look for a Doodle poll for a time we could regularly meet to discuss progress.
Details follow. Thanks in advance for your help in making this a success!
Rob
---
Organizers. Here's our team:
- Gary Grider (LANL)
- Scott Klasky (ORNL)
- Glenn Lockwood (LBNL)
- Kathryn Mohror (LLNL)
- Lee Ward (SNL)
Lucy Nowell is our ASCR contact, and I expect that Laura Biven will be included and possibly take over as contact before we are done.
We have a mailing list: ssio-2018(a)lists.mcs.anl.gov<mailto:[email protected]> for discussion. You have been added to the list. Please let me know ASAP if for some reason you cannot help.
Topical areas and initial discussion. Five focus areas were discussed for further investigation at the prior workshop.
- HW/SW architectures
- Metadata, name spaces, and provenance
- Supporting science data
- Integrating with external services
- Understanding SSIO systems
I propose that we do a "virtual" summit (i.e., videoconference) for 3/4th of a day, rather than flying somewhere for a day to meet in person. Each organizer would present their view on challenges as relate to mission and discuss if/how they think we should adjust the topical areas in response to changes since 2015. I think of these as 20 minute presentations that are devoid of our own research specifics/propaganda and focus on these big picture items. From this we would select a set of topical areas used to organize the factual status document (see below).
Factual status document. We are modeling our process after the BES "basic research needs" approach (see attachment), with adaptations to the scope of the area and ASCR needs. This process includes the generation of a factual status document (FSD) prior to the workshop. There is a separate group working down this path for an Extreme Heterogeneity FSD and workshop, too. The FSD can be thought of as the background sections of the final workshop document, and that's how we will approach it.
With this in mind, we will organize the document around our initial guesses at topical areas, understanding that the topical areas may have to be adjusted in the final report as a result of discussion. The goal would be to complete the following for the FSD:
- Introduction
- Science Challenges
- Technology Challenges
- Knowledge Gaps -- what we need to know that we don't know
- Research Needs
- Topic 1
- Definition/summary of area
- Current status
- Topic 2 ...
- Appendix: References
This is likely to be a 50 page document.
This document would be partitioned into sections to be managed by individual organizers. Rather than requesting white papers from potential attendees, organizers would invite experts in respective areas to participate in the development of requisite sections. It might be easiest to develop this document using an online system such as Google Docs, understanding that eventually it would be turned into a Word document and finally into a PDF for publication as the workshop report.
Position papers. In the Extreme Heterogeneity planning, we asked for position papers. We got over 100 of them. Parsing them has been a real challenge for the organizers. We should think about what approach we might employ for gathering input from the community prior to the workshop and what we would do with that material prior to the workshop vs. incorporating into the final report.
Workshop. We will shoot for around 60 attendees, maybe 30/20/10 lab/university/industry. This helps keep the justification and planning fairly simple. We would plan for a 1.75-day workshop to facilitate covering five topical areas as separate, single-track discussions. If for some reason we end up needing only 1.5 days for the meeting proper, this would leave more time for the organizers to meet and discuss report completion.
We will (mostly) assume that attendees have read the FSD prior to arrival at the meeting. Never the less, we will spend 1/2 day bringing the attendees up to speed, focusing on the background material from the FSD. This is similar to the opening plenary session portion of the BES workshop structure.
The remainder of the meeting could follow the model from the 2014 SSIO meeting: 90-minute moderated discussion sessions, with notes captured live in PPT. With 1.75 days we can accommodate 5 total discussion topics. We will place a topic of broad interest last to encourage attendees to participate for the full meeting. These sessions will identify current status in the area, its CS challenges, its opportunities, and its research needs, reflecting the breakout session portion of the BES workshop structure. Because the meeting is single track, there is no need for outbrief sessions (i.e., the participants will have been present for all discussions).
Report. The final report will be generated by augmenting the FSD with material summarizing the outcomes of the workshop. We will target a 75-100 page document (the 2015 report was approximately 90 pages). The outline for the report will look something like this:
- Executive Summary
- Introduction
- Science Challenges
- Technology Challenges
- Knowledge Gaps (need to understand what this typically includes)
- Research Needs
- Topic 1
- Definition/summary of area
- Current status
- CS challenges and opportunities
- Priority research directions (3-5)
- Topic 2 ...
- Priority research directions
