A column based in-memory database with a url compatible query language and very small response times. Primary development aim was a server capable of calculating histograms over single columns accepting any number of filter queries on any attribute (column) of the dataset. The original use case was to allow an interactive analysis of web logs.
- Configure dummy data source (jdbc based) in the default.yml config file. This data will be loaded into memory, make sure you choose something small enough to fit into your RAM.
- Run it with
gotcount-service.jar server default.yml - Open
localhost:8080to access the service. The server will return path options if available
Append ?query= to any data returning url to define a filter. attribute=1 will filter on the column attribute with the value 1. Add more attribute filters by using the semicolon devider ;.
Query := Filter(';' Filter)*
Filter := column ':' Action
Action := Value | UnaryOperation | BinaryOperation | SetOperation
UnaryOperation := ( '!' | '>' | '<' ) Value
BinaryOperation := Value '-' Value
SetOperation := ( 'IN' | '!IN' ) ( Value | ( '[' | '(' ) Value (',' Value)* ( ')' | ']' ) )
Value := Number | String
gotcount is split into four projects:
- *-core: the database itself
- *-parser: the query parser
- *-query: the query atoms
- *-service: the web/rest service
gotcount makes heavy use Compressed Bitmaps on the data layer and uses parboiled for query parsing. The service component is build using Dropwizard.