Scylla: Full Text Search Integration

[tzach]

Cassandra integrate with tools like Lucene / Solr / ES
Similar integration with Scylla will be useful for many users.

[ddorian]

Do you have any plans on tackling this ?

Meaning implementing another search-lib (is there any that work with async writes on xfs?)
or implement something external like datastax-solr
or implement something different external like https://github.com/vroyer/elassandra
or do your own inverted-index-lib
or something else?

[gkze]

This would be interesting to see

[ghost]

Why not the possibility to use Sphinx ?
It's written in C++.

[ddorian]

New inverted index libraries have come up:

https://github.com/phaistos-networks/Trinity
https://github.com/tantivy-search/tantivy

Scylladb can integrate with one of them and do something like elassandra .

Still, the best way would be what couchbase does, which is separate nodes by their type (for most efficient hardware uses). So you would have 1 cluster where some nodes are in-memory-pedis (ram, ok-disk, ok-cpu), some are search-engine (ram+cpu), some are columnar-store (hdd+cpu), some are scylladb (ssd+ram+cpu). Some type of queue between them and some type of spark sparkle on top. (don't know if it makes sense + needs alot dev-time)

[fulmicoton]

I'm available if you have questions about tantivy.

[didip]

Sphinx on Scylladb would be glorious. Especially if it's as easy to use as CREATE CUSTOM INDEX.

Similar to how this project does it: https://github.com/Stratio/cassandra-lucene-index

[Remember2015]

I'm available if you have questions about tantivy.

So is there any solution or possible way to integrate tantivy / sphix with scylla?

[dorlaor]

We haven't look at it yet, you're welcome to give it a try

On Sat, Dec 22, 2018 at 5:10 AM 子龙 [e-mail] wrote:

I'm available if you have questions about tantivy.

So is there any solution or possible way to integrate tantivy / sphix with
scylla?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
https://github.com/scylladb/scylla/issues/1083#issuecomment-449569037,
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABp6RSIfimyisxAb47A3_JmefRTZYYqJks5u7i8ygaJpZM4H2xW9
.

[elgalu]

It seems for now one needs to ingest the data into both Scylla and Elasticsearch using dual writes and then perform the textual search:
https://www.scylladb.com/2019/03/07/scylla-and-elasticsearch-part-two-practical-examples-to-support-full-text-search-workloads/

[dorlaor]

Right, Scylla doesn't have full text search, we haven't gotten to it..
We do have a LIKE operator which runs a scan on several partitions
and can retrieve a regular expression match. It's not an index but can be
a poor man's replacement to search in a relatively small range/partition.

On Wed, Jul 8, 2020 at 10:31 PM Leo Gallucci [e-mail]
wrote:

It seems for now one needs to ingest the data into both Scylla and
Elasticsearch using dual writes and then perform the textual search:

https://www.scylladb.com/2019/03/07/scylla-and-elasticsearch-part-two-practical-examples-to-support-full-text-search-workloads/

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/scylladb/scylla/issues/1083#issuecomment-655910300,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AANHURLUKZ6KYULPXFK5W2DR2VI3VANCNFSM4B63CW6Q
.