Hi everyone! We're using the StarRocks K8s operator for deploying the cluster (in shared data mode) and wondering about the recommended deployment sizing: For production scale workload, given the same resources, is it usually recommended to deploy many lightweight CN pods or fewer more powerful CN pods? For example: 20 CN pods with 64GB RAM and 16 CPU cores, vs 10 CN pods with 128GB RAM and 32 CPU cores. What is the best practice?

both should be ok

Thanks @jeff.ding

@Rotem Adhoh We've found that bigger nodes with more RAM work better for us. We'd get constant OOM errors when running large queries when we had nodes with less RAM. @jeff.ding is that expected behavior?