Originally from the User Slack
@Sh4d1: Hello! I was wondering if there is a way to horizontally scale the CDC driver (rust one in my case) ?
@Krasimir_Popov: What exactly is your use case? Are you trying to capture events from scylla to do actions in external system or something else?
@Sh4d1: Yes we’re currently capturing all inserts to a redpanda (Kafka) cluster
And I’m wondering if it’s possible to horizontally scale the CDC driver
@Krasimir_Popov: Theoretically you can, but I never tried such a thing. You need to take into consideration something like Token Range Sharding. ScyllaDB’s CDC data is partitioned by token ranges. Distribute these ranges across driver instances to parallelize ingestion. Each instance of your event consumer (rust microservice pod)will follow a different range.
Maybe start small and try to split the range by 2 statically hardcoded 
and if it works you can keep hacking to make it capable to scale up and down exchanging who(which instance of the rust pod) is following what range via etcd, redis, consul or special scylladb table
@Sh4d1: Yeah that was kind of what I had in mind! Thanks 
@Krasimir_Popov: Let me know if the hardcoded approach works, this is actually a good idea for an open source project to easy grab some github stars
I am not very good with rust but I can definitely make something on top of the go driver
@Sh4d1: Won’t have a lot of time to in the near future, but I’ll definitely need this at some point I’ll try to find some time to hack something up on top of the rust driver (if I manage )