Talks this Week: Open Data at ESTC in Vienna and TOP-IX Annual Conference in Turin

DECEMBER 2, 2010

This Thursday I’m in Vienna to participate in the Linked Open Data workshop at ESTC 2010 talking about open data on the semantic web (putting the ‘open’ in linked open data!). Update: slides from talk at ESTC.

Then on Friday (3rd Dec) I’m at the TOP-IX Open Data conference in Torino, Italy to speak on Public Sector Information and Open Data (on the 2nd fellow OKFNer Friedrich Lindenberg will be coordinating a CKAN hack day in Turin as part of the run-up to the main event). Update: slides from TOP-IX talk.

Notes from ESTC

Live notes from LOD session (no guarantee of accuracy or completeness)

  • - lod cache
  • 18 billion triples
  • 8 x 8 core, 16GB servers
  • 350m geolocations

[ed]: What’s the cost (if you weren’t corporate)? On, e.g. EC2 (using high memory reserved extra large), this is 8 * ~3k = 24k per year.

Reserve 16GB RAM for every billion triples

Column wise compressed storage architecture

  • Columns better for analytics
  • Rows betters for lots of transactions
  • Virtuoso moves to column based architecture

Monetdb - what can we learn from them

NoSQL - in some ways a cultural thing, a rebellion against the authority of SQL

  • Get rid of (some parts) of ACID
  • Because google were doing it people thought they should too
  • Something of a reaction now