Hi all,
I'm managing since 3 years now an high-availability Ceph clusters for
our virtualization infrastructure (Proxmox VE). We use Jewel with rbd.
It works perfectly well and meets our data integrity and performance needs.
In parallel, we want to add a new Ceph cluster for data storage with a
rather high read/write usage, using pools configured in CephFS.
Here is the configuration I plan to set up in our infrastructure spread
over 3 separate buildings (linked by a 10Gbs fibre network) :
In each building, 1 MON/MDS server and 2 Ceph OSDs servers (either 3
MON/MDS servers and 6 Ceph OSDs servers in total) with the following
characteristics :
MON/MDS server (all 3 identical) :
model-like DELL R640, 64GB RAM, 2xXeon@2,4 GHz, 2xM2 240GB (system),
2xSSD Mix-Use 480GB, dual port 10Gbs SFP+
Ceph OSDs server (all 6 identical) :
model-like DELL R740xd, 64GB RAM, 2xXeon Scalable Family, 2xM2 240GB
(system), 24xSAS 10k 1,8TB, dual port 10Gbs SFP+
Overall, according to the above requirements (notably the use of
CephFS), does this configuration seem correct ?
Also, I have two questions regarding two specific points :
1) Is it relevant to put both MON and MDS on the same physical machine
(considering that there are in total three physical machines for that
purpose) ?
2) As I'm going to use Bluestore, I wonder if it would be optimized to
add 2 SSD disks on each OSDs servers in order to separate WAL (Write
Intensive SSD with low capacity) and DB (Mix Use SSD with high capacity)
from the HDD disks ?
(That would mean on my servers put 22 HDD + 2 SSD instead of 24 HDD)
Thanks for all your suggestions,
Hervé
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
