Datomic

Datomic

• 1.
  Datomic JP Erkelens @jperkelens Http://github.com/jperkelens
• 2.
  Disclaimer!! Nearly everything thatfollows is cribbed from the datomic website: www.datomic.com
• 3.
  Rationale  Database design conventions were created when memory and disk space was limited and expensive  Place-oriented programming  Records should not be erased when new ones are made
• 4.
  Classic DB Architecture
• 5.
  Datomic Architecture
• 6.
  Peers  Library embedded in app  Submits transactions to transactor  Handles query/caching/data access
• 7.
  Transactors  Process transactions serially  Transmit changes  Index in background
• 8.
  Consequences  Reads never lock out writes  Each peer has own cache  Higher memory footprint/quicker responses
• 9.
  Data Model  Data is immutable  How? Time.  Datom  Consists of entity/attr/value/transaction  We don't update records or documents  We add/remove datoms  Minimal Schema
• 10.
  Programming Model -Queries  The Peer pulls data from storage as needed and caches  It receives updates from the transactor  Queries run from a merged view of the two  After a while minimal network activity is needed  No strings! (Maps and vectors)  Query language – Datalog.  Implicit joins
• 11.
  Programming Model –Time  Apps always work on a consistent snapshot of the database  Queries are applied to values of the database in time, or windows
• 12.
  DEEEEEMMMOOOO!!! Additional Resources  Www.datomic.com  Http://github.com/limadelic/datomic  https://github.com/gns24/pydatomic