Hi, All. In my opinion, it’s quite challenging to maintain the same level of latency while reducing costs. Fortunately, our use cases can generally be divided into two categories: for key business scenarios, we tend to prioritize low latency over cost; for non-core scenarios (such as log or metric collection), we tend to prioritize cost over latency.
Therefore, if the implementation can support switching between these two types of topics (disk less topic <-> traditional topic/tiered-storage topic), that would be perfect for us. We could then split the Kafka clusters to host different types of workloads according to our objectives. Regards Jian -- Regards Fu.Jian ------------------------------ Cisco Communications, Inc.
