On 22/07/15 20:39, Shneur Zalman Mattern wrote:
Third test:
We wanted to try CephFS, because our client is familiar with
Lustre, that's very near to CephFS capabilities:
1. I've used my CEPH nodes in the client's role. I've
mounted CephFS on one of nodes, and ran dd with bs=1M ...
- I've got wonderful write performance ~ 1.1
GBytes/s (really near to 10Gbit network throughput)
2. I've connected CentOS client to 10gig public network,
mounted CephFS, but ...
- It was just ~ 250 MBytes/s
3. I've connected Ubuntu client (non-ceph member) to 10gig
public network, mounted CephFS, and ...
- It was also ~ 260 MBytes/s
Now I have to know: perhaps ceph-members-nodes have
privileged access ???
While you're benchmarking, it's a good idea to try both the kernel
client and the fuse client. You may find one works better than the
other, and we'll find the numbers interesting too.
You're using giant, the latest LTS release is hammer -- if you're
interested in cephfs you'll be better off with hammer (lots of new stuff
going in all the time).
Aside from that, it's kind of surprising that your servers are working
as better clients than your clients. Do the clients definitely have all
the same kernel+ceph versions as the servers? Is the link between the
clients and the servers definitely 10G all the way across your network?
Is a pure network benchmark seeing the full 10gbps between a client node
and a server node?
Cheers,
John
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
