.. meta::
:keywords: DataONE, CCIT, 20110330, VTC
2011-03-30 - DataONE Developer Call
===================================
:Attendees: Dave Vieglais, Paul Allen, Rebecca Koskela, Bruce Wilson, Ryan Scherle, John Kunze, Robert Sandusky, Mark Servilla, Rob, John Cobb, Matt Jones, Chris Jones, Roger Dahl, Nicholas Dexter
Agenda and Notes
----------------
**1. Demonstration screen casts from the review**
The demo screen casts are located in subversion at:
EVA demo (R script): https://repository.dataone.org/software/allsoftware/demos/eva_demo_1/screencast/DataONE-species-prediction-analysis.mov
Morpho demo: https://repository.dataone.org/software/allsoftware/demos/morpho_upload_1/20110217_cjones_morpho_d1_demo.mov
Mercury demo: https://repository.dataone.org/software/allsoftware/demos/mercury_demo_1/DataONE-Mercury-general-facets.mov
FUSE file system demo: https://repository.dataone.org/software/allsoftware/demos/fs_demo_1/DataONE-FS-v0.2.mov
slides: https://repository.dataone.org/software/allsoftware/demos/
- General impressions
- voice over important.
- Applicability of each
- Subtitles only vs voice over + subtitles?
- Refactoring?
- Distribution preferences (vimeo, youtube, ...)
- Distribution restrictions?
- Consistency between screen casts?
Paul - send out ASAP, as is. Make them public.
Bob - make them fully accessible - so voice over and transcripts are important. stop/start controls. Need to add introductions and timing steps to provide guidance through the screen casts.
Resolution 1280x720 (largest handled by youtube)
Sandusky notes below [0]
Probably best to have a non-tech person write the transcript to provide a broader reach.
Next steps:
- discuss at LT
- power point slides for steps
- transcripts
- voice over
**2. Data replication constraints (continued)**
Bandwidth performance between each of the CNs and to a client.
See http://mule1.dataone.org/ArchitectureDocs-current/notes/time_bandwidth_constraints.html
- add periodic tests for bandwidth between nodes as part of the monitoring infrastructure
- also useful to add in "save to disk" times
- also add n clients
- check the routes that are being used
- perfsonar (being used by internet2) - could also test nodes in between
- iperf
- need to look at latency in addition to throughput
- also add in checks for member nodes (largest data holdings)
ref: TeraGrid metrics:
Speedpage (NxN test of GridFTP performance between TG sites) <http://speedpage.psc.teragrid.org/speedpage/www/speedpage.php>
WE also have a iperf infrastructure that does igbps tests for speed and jitter, but it is mostly turned off during most times now since the network works well at 1 gbps adn we do not want to test at 10 gbps becuase it impacts production work.
Perfsonar link: http://www.perfsonar.net/ Martin Swany (U. Del) is the main force behind this and it is catching hold in internet2 and ESnet and elsewhere. The idea is to develop a set of non-intrusive tests that give good correlation to netowrk performance under high loads.
**3. Any other business**
- CCIT Meeting Scheduled for 19-21 July. Any preparations that need to be
checked on?
- Matt will send out a reminder email for the Doodle poll (need a head count)
- DataONE Users Group meeting July 11-12
-- Meeting likely to focus on ITK (December 2010 meeting focused on MNs)
-- Perhaps have some ITK development partners attend?
- perhaps invite development partners such as microsoft?
- Matt, Bob, John will be attending
- Use meeting as a forum for gathering requirements, perhaps usability testing, etc.
- Need strong presentation about what the ITK is
- Use UG to help prioritize the tool selection and features to be supported, etc
- What are some important questions that need to be addressed as outcomes from the DUG meeting?
- e.g. what are the three tools that should have dev focus?
- other groups that can be convinced to help with development (of itk pieces)?
- Finalising the agenda - time frame? Draft agenda approx end of May.
- Invitee list - needs to be broad cross-section to minimize bias in prioritization. Input on invites from CCIT would be very helpful
- approx 50 attendees
[0] Sandusky notes
Overall, the screencasts are quite good.
General comments:
- accessibility: these are OK for sighted people, but not for those with impaired vision. Recommend (1) voice overs and matching transcripts.
- accessibility: controls are important for people with a range of cognitive abilities
- add outlines at beginnings 1- step A 2- step B 3- step C and insert frames along the way to help keep the viewer oriented
- can these screencasts be put into a contextual frame? E.g., "Tools for scientists being developed to work with DataONE" on dataone.org
Potential applications:
- push out to WG membership and potential members
- push out to DUG membership and potential members
- add to DataONE public Web site
- push out to other potential ITK toolmakers
ITK: Morpho Upload Demonstration
- Needs a little more context, which could be provided by a voice-over.
- Might be better with pauses between slides with user start-stop control.
- Second part, starting at 2:32, with title "KNB data synchronize to DataONE" and then showing Mercury and retrieval really needs more contextual information.
What tool was used to create this screencast?
ITK: Species Distribution Model in R
- Pointer bouces around a lot ca. 00:20 - 00:27; is there some other way to highlight screen regions of interest?
- Interstitial titles might help orient the viewer; show the N steps covered in the screencast, highlighting the step that's about to be shown
- Some typos in results returned from Mercury (1) "dat apackage" instead of "data package"; (2) "speccies" instead of "species"
ITK: Facetd Search of DataONE Using Mercury
- pace on this one was better overall than previous two
- the COinS stuff at end went a bit fast perhaps
ITK: DataONE as a Tagged File System
- pace on this one was generally good
- ca, 00:15; consistency: on-screen, data_providers but title says data sources (maybe this is OK)
- ca, 2:50; section on "describes" lost me the first time through
- ca, 3:50; typo (1) "as as" instead of "as" in the title
=====
[4:21 PM] Bob: what tool was used to create the screencasts?
[4:30 PM] John: do we have a sense of what is the "needed" BW?
[4:30 PM] John: to compare with this experience?
[4:31 PM] John: Are we doing periodici iperf connection tests between nodes?
[4:32 PM] Matt: dd if=/dev/urandom of=test_100M.bin bs=1048576 count=100
[4:51 PM] John: I may be there
[4:58 PM] John: and even a university admin-critter might be useful in small doses
[5:00 PM] Bob: community building!