RediSearch open source cluster support?

[Άγγελος Δημόπουλος]

Hello,

As far as i have read cluster support is only available as part of Redis Labs Enterprise. What if a Redis cluster is set up according to this https://redis.io/topics/cluster-tutorial and RediSearch is loaded to each master?

I mean what will be the behavior, it will not work at all? It will work like it was a standalone mode? Are results not guaranteed?

Also are only search indexes affected, TEXT fields? If a document with only GEO, TAG and NUMERIC fields is created, would it work as expected?

Thank you in advance.

[Guy Korland]

It should work, but a single index will not scale out or distributed across the cluster.

But, multiple indexes will be spread across the cluster.

[Άγγελος Δημόπουλος]

So in order to see if i got this right.

Creating an index FT.CREATE idx SCHEMA characteristics TEXT SORTABLE location GEO myTag TAG SORTABLE and then adding multiple documents to index idx.

All this will be stored to the same master node and then a query on this index will be processed only on this node?

[Meir Shpilraien]

Yes but notice that you might get keys created on shards they are not suppose to, to work around it I would decide on unique id for each index and add this unique id to the index name and to all the documents. Example:
idx{uuid} for index name
doc{uuid} for documents ids

Also you are going to need to develop a smart client that knows to send the ft.add, ft.create, and ft.search commands to the right shard ...

[Άγγελος Δημόπουλος]

Ok I understand, but will i really get any benefit as far as the query will not be distributed.

My use case is that i want only a single index with geo and text fields in order to do some metrics and benchmarks compared to mongodb for example on "spatiotextual" queries. But the requirement is that it has to be a cluster setup.

Can i somehow benefit from the cluster with opensource Redis or my only way to achive this is Redis Enterprise? :/

[Άγγελος Δημόπουλος]

Any info about this?

Thank you.