What is your object size will be? 70T RAW is such small, I think is better add hardware to your RBD cluster and run object service here k Sent from my iPhone > On 8 Jul 2021, at 14:17, gustavo panizzo <gfa+ceph@xxxxxxxxxxxx> wrote: > > Hello > > I have some experience with RBD clusters (for use with KVM/libvirt) but > now I'm building my first cluster to use with RGW. > > The RGW cluster size will be around 70T RAW, current RBD cluster(s) > are in similar (or smaller) size. I'll be deploying Octopus > > Since most of the tunning is pretty different (big number of PGs, > bluestore_compression_*, bluestore_min_alloc_size_*) I wonder if > makes sense to run both workloads in the same cluster or if it would > be better to have dedicated clusters. > > However bigger clusters (AFAIK) are more stable, what is other people doing? > Single cluster for all workloads or a cluster per workload? > > thanks! > > PS: Asking for the future, what about CepFS? should it share cluster? > > > -- > IRC: gfa > GPG: 0x27263FA42553615F904A7EBE2A40A2ECB8DAD8D5 > OLD GPG: 0x44BB1BA79F6C6333 > _______________________________________________ > ceph-users mailing list -- ceph-users@xxxxxxx > To unsubscribe send an email to ceph-users-leave@xxxxxxx _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |