teragrid quarterly meeting dec 5 - 7, 2006 data, visualization and scheduling (dvs) update kelly...
TRANSCRIPT
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data, Visualization and Scheduling (DVS) Update
Kelly Gaither, DVS Area Director
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data Movement
The alpha draft of the data toolkit has been completed and has been initially reviewed by Lee: Needs some tweaking to make it complete and consistent with
other toolkit definitions Will be distributed to the data working group after that
Key items to focus on over the next several weeks: Formal testing plan to ensure that deployed tools are tested and
stay working (configuration and change management plan) Complete analysis of data transfer speeds
• PSC Speedpage (http://gridinfo.psc.edu/gridftp/speedpage.php) has the raw numbers for point to point transfer times. Group at PSC will be investigating and documenting bottlenecks, and what can be reasonably expected given the current infrastructure.
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data Management
Phil Andrews is leading discussions about globally available file systems going forward. Current examples of this are GPFS-WAN and Lustre-WAN
Will be looking at viability of Amazon S3 storage for TG user community
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data Workshop
January 9-11, San DiegoThe outcome will be a draft of a TG wide data
movement/data management plan going forward
If you are attending: Please email Mark Sheddon to register Please make your hotel reservations by the end
of this week
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data Collections
Data Collections RAT Report is complete Defined what constitutes a formally designated TG data
collection: Classes:
• Research data collection• Resource or community data collection• Reference data collection
Types:• Type 1: Generally Accessible Data Collections
– Satisfied basic requirements• Type 2: Compute Associated Data Collections
– Requires routine usage of TG compute or visualization resources to process data (e.g., Purdue Environmental Data Portal consolidates several earth observation data collections)
• Type 3: Globally Available Data Collections– Large data collections available on global file systems (e.g., NVO)
• Type 4: TG Affiliated Data Collections– Data collections demonstrating a link to TG, for example a
demonstrated TG user community desiring access
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Data Collections
Data Collections Working Group: Charter is complete Working group is expected to begin
meeting beginning of 2007Going forward in 2007 we expect
folks to approach TG for inclusion as a TG data collection: Will be completing a process for
becoming a TG data collection before the review
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Visualization
A half-day visualization tutorial “Remote/Collaborative TeraScale Visualization on the TeraGrid,” was taught at IEEE Visualization on October 29, 2006: Taught by Kelly Gaither (TACC), Mike Papka (ANL),
Joe Insley (ANL) and David Ebert (Purdue) Visualization Workshop for Users:
Coordinating a full day workshop on the Monday of TG ’07 (June 4, 2007)
Limiting enrollment to 25-30 maximum Picking ~8 power users to begin working with now
and visualize their data as examples for the workshop Will be distributing a call for participation by Jan, 2007
TeraGrid Quarterly MeetingDec 5 - 7, 2006
TeraGrid Visualization Gateway
Q4FY06 Accomplishments Presentations & demonstrations of TeraGrid Visualization beta portal at SC06 Collaborative and Remote Visualization Functionality
• Remote Paraview on UC/ANL systems• Remote & Collaborative Visualization on Maverick
Offers support for TeraGrid User Portal accounts and community accounts• Current TeraGrid User Portal users can login with their TGUP account and have
full access to Viz Gateway.• Community users can create their own accounts with restricted access.
Q1FY07 & Future Plans Continue development to production quality portal Integrate Additional Visualization Tools
• Expand current set of visualization tools and work with other RP sites to see what they can offer (e.g., Purdue has expressed interest in including a Maya portlet).
• Look into registering ‘visualization services’ with the portal, either through web services or importing additional functionality.
Milestones:• Have portal v1.0 complete and in production by TG07 Conference (and TG Viz
User’s Workshop)• Paper submission to TG 07 Conference
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Remote Visualization Portlet
TeraGrid Quarterly MeetingDec 5 - 7, 2006
ParaView Portlet
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Scheduling
Co-scheduling/metascheduling RAT: Leadership and Members
• Warren Smith (TACC) and Patricia Kovatch (SDSC)• Members from each RP site
Accomplishments to Date: Performed a user survey, summarized and discussed the
results Performed an RP survey, summarized and discussed the
results Performed a review of metascheduling tools, summarized and
discussed the results Drafted a set of recommendations Finishing final report (1/5/2007) Developing a scheduling working group charter
TeraGrid Quarterly MeetingDec 5 - 7, 2006
Scheduling
RAT Recommendation Topics: Advance Reservations Co-scheduling On-demand Scheduling
• Highest priority
• Preemption Automatic Resource Selection Ensemble Workflow
Four Primary Tasks: Update user metascheduling requirements Compile information about scheduling environments on RP systems as
well as RP requirements and preferences in the area of metascheduling Identify and perform a paper evaluation of metascheduling tools Develop recommendations for metascheduling in TG