Hi,
I've been reading this blog post [1] about how works, it says that
> Rows are freed starting from the least recently used ones, with insertion counting as a use ..
Is there any way of tuning this? so counting only the reads.
Correct me if I'm wrong but, for workloads where you have a continuous
flow of writes for different keys but only a small subset of them are
often used at reading time and are also not frequently updated, the
fact of counting any individual write would reduce the hit ratio
level, making this feature useless in some scenarios.
There is an enable_cache option in scylla.yaml.
--
You received this message because you are subscribed to the Google Groups "ScyllaDB users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scylladb-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scylladb-users/30152c0a-dfe2-4033-9dbc-7e18d6066f55%40googlegroups.com.