What is the relation/difference between Amazon DynamoDB and FlockDB?

Question

I am trying to implement incremental PageRank using Monte Carlo Sampling. My data includes millions of relational data between 20 different entities.

I found Amazon DynamoDB and FlockDB to be two good candidates for my task as data storage. A brief comparison is here. Can anyone point out which one is more suitable for my PageRank task?

Answer 1

This answer should be taken with a healthy dosage of salt because I'm not 100% about your requirements.

First and most obviously, this paper is from Twitter (Personilised page ran - SALSA). I know at some point the head of their recommendations / personalisation said they used it with a bunch of other techniques (This is from Oct'11) https://fbcdn-sphotos-b-a.akamaihd.net/hphotos-ak-snc7/300267_581958426587_2140504611_n.jpg

In terms of which is more suitable for your task, I would go with FlockDB. The primary reasons are that it supports graph based queries which are going to be quite common in your code. It also has some explicit persistent graph functions such as persisting edge information.