Hello all,
are there any ideas or approaches to use Vector Searches in Scylladb?
This should be supported with Cassandra 4.0 and I think it would be a nice feature
to support similarity searches based on text embeddings.
Hello all,
are there any ideas or approaches to use Vector Searches in Scylladb?
This should be supported with Cassandra 4.0 and I think it would be a nice feature
to support similarity searches based on text embeddings.
Hi Jeremy,
ScyllaDB doesn’t support vector searches. One possible option to include ScyllaDB in a vector search project is to use a tool like Vald that can do vector search and use ScyllaDB as backup.
Hi Attila thanks for your response.
I didn’t know about Vald looks very interesting I just know about the typical vector dbs like redis, pinecon, cozodb.
Maybe somebody from the Scylladb Core team can give a outlook
You might also find this tutorial useful.
It explains the basics of Machine Learning (ML) feature stores, and it explains how ScyllaDB can be a critical part of your feature store architecture.
I know it’s been a while since your original post, but I was curious—how did you end up tackling the initial need you mentioned? Would love to hear how things have been working out.
Happy to connect for a quick chat if you’re up for it!
Thanks,
Jakub
Since the original post, there’s been an update on vector search in ScyllaDB.
Vector search functionality is actively being worked on and is under development in ScyllaDB.
The expected CQL syntax includes:
Defining a vector column:
ALTER TABLE cycling.comments_vs ADD comment_vector VECTOR <FLOAT, 5>
Creating a vector index:
CREATE INDEX IF NOT EXISTS ann_index ON vsearch.com(item_vector) USING 'usearch'
WITH OPTIONS = { 'similarity_function': 'DOT_PRODUCT' };
Performing vector similarity queries::
SELECT * FROM cycling.comments_vs
ORDER BY comment_vector ANN OF [0.15, 0.1, 0.1, 0.35, 0.55] LIMIT 3;
With this feature, ScyllaDB will enable a wide range of use cases, including:
We’ll share updates with the community as this feature becomes available.
Learn more!