CODEWITHSUNDEEP
nosqlcassandradata-modeling
Jhon
Jhon

Reputation: 433

Cassandra datamodel

I try to build a backend for quiz game statistics : I want to know recognition percent for each image globally and recognition percent of images by country.

Currently datamodel is : 


TABLE results (
  country text,
  percent float,
  image_id text,
  fail int,
  ok int,
  shown int,
  PRIMARY KEY (country, percent, image_id)
);

I can query results by country, but I can't get global results(images with higher percent).

Any hint ?

Upvotes: 2

Views: 125

Answers (2)

Serge
Serge

Reputation: 38

You need two CFs.

Another way is to use ordering partitioning and use keys of CompoteType, so your key is not PRIMARY KEY (country, percent, image_id) but PRIMARY KEY ((country, percent, image_id)).

You also need another column, image_id and you could put a secondary index on that image_id so that you can do queries where image_id=x

Upvotes: 0

Wildfire
Wildfire

Reputation: 6418

It's possible to define "TOTAL" country that will hold sum of stats for all countries.

P.S. Making percent part of primary key doesn't seem to be a good idea due to eventual consistency in cassandra: it won't be possible to do an atomic update of percent field.

I think there are two possible situations: either the data set would be small enough - then any SQL DB with secondary index on the percent field will do job better than cassandra, or there would be too much data for SQL DB - then Cassandra with this data model won't work either, only the map/reduce of partial counts will do.

Upvotes: 2

Related Questions