Goals

1

Welcome!
2
Goals
3
XLDB Goals

1. Identify trends, commonalities and major
   roadblocks related to building extremely large
   databases
2. Bridge the gap between users trying to build
   extremely large databases and database solution
   providers worldwide
3. Facilitate development and growth of practical
   technologies for extremely large data stores

4
10/2007

1. Identify trends and major roadblocks related to
   building extremely large databases
2. Bridge the gap between users trying to build
   extremely large databases and database vendors
3. Understand if and how open source projects like
   the LSST Database can contribute to the previous
   two goals in the next few years

5
09/2008

1. Continue to understand major roadblocks related
   to extremely large databases with an emphasis on
   complex analytics
2. Continue bridging the gaps within the XLDB
   community including science, industry, database
   researchers and vendors
3. Build the open source SciDB community

6
08/2009

1. Reach out to the XLDB communities outside of the
   USA
2. Connect with more science disciplines and
   communities which were underrepresented in the
   past workshops
3. Review existing XLDB engines and solutions and
   discuss how to move the state of the art forward

7
Some Highlights From Past Workshops
8
Many Commonalities in Data Analytics

• Pattern discovery
• Outlier detection
• Multi-point correlations in space and time
• Multi-d aggregation
• Unpredictable query load

9
Many Common Trends

• Rapidly increasing size
• Size growth is increasing
• Discarding valuable data
• and complexity
• Data structures and techniques applied more
  complex
• Capturing conditions that cannot be reconstructed
• and flexibility needs
• Rapidly changing or unknown requirements
• Abandoning normalized schemas
• Platforms
• Shared nothing parallel architectures on
  commodity clusters
• Rebuilding, not reusing

10
Roadblocks

• Funding
• Disconnects
• Especially vendor-users, science-academia
• Analytical tools not keeping pace
• Lack of fault-tolerant, scalable and affordable
  tools
• SQL APIs - set orientation and low-level
  interfaces, poor integration with analytical
  tools and procedural languages

11
XLDB Ignited Initiatives

• SciDB
• Science Benchmark

12
Things We Decided_at_ XLDB2
13
?

• SciDB
• Disconnect from XLDB
• Publish collected use cases
• Reach out to more sciences
• Periodically inform XLDB community

14
?

• Science challenge
• Try to define a standard challenge focused on
  data-intensive scientific queries

15
?

• Wiki
• Make more visible and publicize
• Attempt to recruit an active moderator

16
?

• Organize XLDB3
• 2 days, around VLDB or SIGMOD
• Reach out to communities in Europe and Asia
• Connect with more science disciplines and
  communities
• Plus, consider tutorial at VLDB or SIGMOD to
  tell larger database community about real
  science requirements

17

18
It Is All About Ad-hoc Discussions

• You are expected to speak up too
• Refrain from sale speeches
• Discussions are not electronically recorded
• Detailed report will be released
• Once OKed by workshop participants

19
Attendance Rough Breakdown
xldb1 xldb2 xldb3
43 41 46 Data-intensive scientific users
21 19 15 Data-intensive industrial users
30 19 24 Vendors, incl. startups
6 21 15 Academia, db research programmers
53 62 52 ?Total count
20
Agenda
http//www-conf.slac.stanford.edu/xldb09/agenda.ht
m
21
Logistics
22
Coffee Breaks / Lunch / Dinner

• Coffee breaks 1030am, 330pm
• Lunch 1230pm
• Reserved space
• Reception 600 pm, dinner 700pm 900pm
• Both at JOLS restaurant
• Special needs just ask
• All free, thanks to our sponsors

23
Reception/Dinner Location

• JOLS restaurant
• Address 283 Avenue Jean Jaurès
• How to get there
• "C1" bus to Part-Dieu (or Brotteaux)
• subway line B direction "GERLAND", take off at
  "DEBOURG"

24
BIG Thanks to Our Sponsors