Thanks Steve. With your approach, I believe it will not allow double
entries for a follower-followee relationship which is good. In our
modeling, a timeuuid field is chosen to be the cluster key and this lets a
time-based ordering, which was one of the design goals I believe.
Actually, I started t
Thanks for your answer Sean. Yes I also viewed the problem as a graph
problem, but then thought it could be an overkilling solution. Because, I
only need to have the followers and followings of a user, nothing more.
2017-03-29 20:35 GMT+03:00 Durity, Sean R :
> This looks more like a problem for
Hi Ferit,
I have implemented the follower/following relationship in a manner similar
to what you describe. However, I can't tell from the description of your
tables if you are using the same convention I used. The two tables I used
were symmetric, as with yours, but with the ids swapped. I inserte
This looks more like a problem for a graph-based model. Have you looked at DSE
Graph as a possibility?
Sean Durity
From: ferit baver elhuseyni [mailto:feritba...@gmail.com]
Sent: Tuesday, March 14, 2017 11:40 AM
To: user@cassandra.apache.org
Subject: results differ on two queries, based on secon
