On 6/21/21 3:19 AM, Eric Robinson wrote:
Suppose you’re building a cluster with 20TB of storage that will run multiple instances of MySQL. Which is better, creating one big drbd disk, or carving it up into multiple smaller ones?
Many small DRBD resources is likely better than one large resource in this case. With multiple resources you'll have independent failure domains as opposed to one large one. Also, each DRBD resource will get it's own set of processes (worker/sender/receiver/etc.) and network socket which will likely help with performance.
You could even use, "options { cpu-mask $mask; }", in your DRBD configurations to ensure each resource gets it's own CPU thread. Here is a link to the relevant section in the DRBD users guide:
https://linbit.com/drbd-user-guide/drbd-guide-9_0-en/#s-latency-tuning-cpu-mask
Also, any thoughts about a preferred filesystem for this scenario?
<snip> Any journaling FS (xfs/ext4) mounted with noatime should be fine. Best Regards, Matt _______________________________________________ Star us on GITHUB: https://github.com/LINBIT drbd-user mailing list drbd-user@lists.linbit.com https://lists.linbit.com/mailman/listinfo/drbd-user
