I have a category of data in my riak application that I can't figure out how to model efficiently. For most of my data, inter-item relationships are relatively compact, and stored solely for the purpose of being provided back to the user as additional information transformed into api and view page links), so I don't need to emulate SQL joins or anything fancy. For these relationships, I'm simply storing the relation directly into the riak object and that works fine. Data is mainly added in batch with the relations and backwards relations precomputed, and data items are very rarely deleted, so i don't have to do lots of updates to change relations.
However, I currently have a data type which has many, many (10k-1mil+) items for a specific relation type, and could potentially have items added and removed from it with relative frequency. I know I'll have other very similar data types soon if I can figure out a way to implement them efficiently. My current model involves using a relational database to store just these relations, but thats the only part of my application that needs a database, so I'd rather ditch the added complexity and find some way to do this in riak. Another ideal property would be to either fetch the whole relation set very quickly (fast-than map/reduce would allow), or to have the ability to efficiently and reliably page through results (sending a million+ items in a single JSON request starts to get pretty unwieldy). Any Ideas? I hope I've stated my problem clearly enough while being generic, I tried to avoid getting domain specific, but I can just lay out the complete domain model if that would be more helpful. Thanks, Alex Thompson www.idigbio.org www.acis.ufl.edu _______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
