[survey] Are you storing documents in RediSearch?

Hi all,

We’d like to know how you’re using RediSearch, and specifically whether you’re storing the documents in it.

Put differently, there are two ways to use RediSearch: mixed and pure. By default RediSearch stores documents in Redis Hashes in addition to indexing them. In this case, RediSearch is both an index and a document store. It also provides the FT.ADDHASH for indexing existing documents.

This behavior can be controlled (i.e. FT.ADD ... NOSAVE ...), which is desirable when the index contains all the relevant data and/or to reduce the memory consumption. In this case, RediSearch is used as a pure index and the document itself is either fully in the index, or stored externally.

So how do you consume RediSearch? Mixed or pure?



If I have understood correctly, this distinction in behaviour would no longer be possible in RS2. In any case, if you look at https://oss.redislabs.com/redisearch/2.0/Commands/#ftadd, there is a badly documented reference to the nosave parameter, which is well documented in previous versions of the module. Moreover, FTADD is a deprecated command now, suggesting that in the future there will be no possibility of using RS as a secondary index, except at a high RAM cost.