as far as I understand Elastic has matured enough by now to be used as a primary data store. Data losses are getting less and less likely and the possibility of freezing indexes introduced in version 6.6 allows for a bigger index to heap ratio which translates to a higher storage capacity for the same price.
Still I'm reading about things like Elassandra (using Apache Cassandra as datastore and Elastic as a secondary index). What are use cases for projects like this given the above (assuming I'm not interested in using Cassandra's query capabilities)? Specifically, I read about three advantages of Elassandra:
- Apache Cassandra's masterless architecture is more stable than Elasticsearch's master-slave-architecture.
- Using Elassandra the same document can be indexed multiple times without using up more space on the disk.
- Elassandra's automatic resharding capabilities (inherited from Apache Cassandra?) allow for easy scaling.
How relevant are these points nowadays and what other (dis)advantages does it have to use Apache Cassandra (or something else) as a primary datastore (and indexing the stored data with Elastic) compared to using Elasticsearch both for storage and for search? For example, what are the differences in performance and disk space needed between these setups?
Thanks for your answers and any sources you can point me to!