Hi. I'm trying to plan hardware for a Nebula cluster deployment. I'm wondering: 1. CPU core count vs. clock speed, which is more important? 2. NebulaGraph Operator vs. bare metal, which is better supported/tested or have better performance? 3. What's the use for multiple disks? Does setting RocksDB to use multiple disks increase performance? Or is it for drive failure resistance? 4. If I'm expecting to have more data than the disk size of the hardware recommendation, do I need to scale up memory and core count accordingly on a linear scale?

According to feedback from frontline implementation engineers, the number of CPU cores is related to query concurrency. There are no reports comparing the performance of NebulaGraph Operator and bare metal. The official performance reports are all based on physical machine testing and do not use the Operator, which is just a method of installation. Multiple disks can increase the speed of data import, not all for resisting faults. A customer previously achieved faster import speeds with multiple disks than with a single disk. The disk size has a related formula, which is multi-linear, so you can proportionally increase or decrease related storage resources linearly.