we have a multi-tenant application, we always filter tenant id first to get the data for a particular tenant then process them. however, 15% of our tenants have 80% of data, so if we use tenant id as the partition key, we will have hot spots. since we always want tenant data in the same cluster, is it ok to use tenant_id as the partition key in this case?
Bringing together the Apache Cassandra experts from the community and DataStax.
Want to learn? Have a question? Want to share your expertise? You are in the right place!
Not sure where to begin? Getting Started
This question was
closed
by
Erick Ramirez for the following reason: The question is answered, right answer was accepted
What is the recommendation for modeling partitions in a multi-tenant application?
If using just the tenant_id as the partition key results in very large partitions for your table, you should add another column in the partition key (in the primary key) to "bucket" your data and make the partitions smaller.
If you have tables with a clustering column like:
PRIMARY KEY ( tenant_id, clustering_col )
you will need to add another column to the partition key like this:
PRIMARY KEY ( (tenant_id, bucket_col), clustering_col )
The primary requirement when modeling data is to design a table for each application query. I mention this because your question is quite specific and answering your question directly doesn't always achieve the optimal outcome you're after.
It's a little difficult for me to provide an informed answer with limited information. If you update your original post with some sample "app questions" (in plain English preferably) so I get an idea of the kind of queries your app is issuing, I'd be happy to update my answer with better guidance. Cheers!
this is good enough for now, let me think more about using multiple columns for the partition key.
Not a problem. Feel free to make a new post if you've got any more questions. Cheers!
5 People are following this question.
