Hello Jiangan, Thank you for the links, they are very helpful. I am wondering whether your Ceph tuning configuration i safe for a production environment.
Thanks -- Howie C. On Thursday, December 12, 2013 at 11:07 PM, jiangang duan wrote: > hope this helpful. > http://software.intel.com/en-us/blogs/2013/10/25/measure-ceph-rbd-performance-in-a-quantitative-way-part-i > http://software.intel.com/en-us/blogs/2013/11/20/measure-ceph-rbd-performance-in-a-quantitative-way-part-ii > > > On Thu, Dec 12, 2013 at 4:22 PM, Wido den Hollander <w...@42on.com > (mailto:w...@42on.com)> wrote: > > On 12/11/2013 09:13 PM, German Anders wrote: > > > Hi to all, > > > > > > I'm new to Ceph and i want to create a Cluster for Production > > > with HP ProLiant DL380p Gen8 servers, the idea is to use 4 of this > > > servers to use as OSD's, and then 3 x HP ProLiant DL320e Gen8 servers > > > for MON. The Data network would be on 10GbE switches and 1Gb the > > > management. Below is the description of each of the servers: > > > > > > > In this case I would run the monitors on the same machines, since a DL320 > > is very overpowered for a monitor in this cluster set up. > > > > > *HP ProLiant DL380p Gen8*: > > > 2 x Intel Xeon E5-2630v2 @2.6Ghz (6-cores) > > > 2 x 64GB RAM > > > 2 x 450GB SAS 15K in RAID-1 configuration for the OS > > > 2 x 100GB SSD in RAID-1 configuration for the Journals > > > > I wouldn't use RAID-1. Short version: SSDs rarely fail and if they fail it > > is due to wearing out. If you use RAID-1 they'll fail at the same moment. > > > > You better use one SSD per 4 OSDs, gives you better performance and > > reliability. > > > > > 8 x 4TB SATA 7.2K to use as 8 x OSD's (32TB raw) > > > 1 x HP Ethernet 10GbE 2-port 530SFP+ Adapter > > > 1 x HP Ethernet 1Gb 2-port 332T Adapter > > > > > > *HP ProLiant DL320e Gen8*: > > > 1 x Intel Xeon E3-1240v2 @3.4Ghz (4-cores) > > > 1 x 32GB RAM > > > > Way to much memory for a monitor. 4GB ~ 8GB is more then enough. > > > > > 2 x 450GB SAS 15K in RAID-1 configuration for the OS > > > 2 x 1.2TB SAS 10K for Logs > > > 1 x HP Ethernet 10GbE 2-port 530SFP+ Adapter > > > > 10Gbit isn't required, but that's up to you. > > > > > 1 x HP Ethernet 1Gb 2-port 332T Adapter > > > > > > > > > I want to know if someone more or less had a similar configuration and > > > to know what are the performance numbers (some benchmarks) for reads and > > > writes, maybe also some iozone or bonnie++ outputs, with several > > > processes (1..10), and different block sizes. > > > Also if anybody had some recommendations or tips regarding the > > > configuration for performance. The filesystem to be used is XFS. > > > > > > > I assume you are going for 3x replication, so with writes you'll have about > > 1/3 of the I/O performance of all the disks. > > > > A 7200RPM disk is capable of about 100 IOps, so that's the figure you > > calculate with. > > > > Ceph performance is very complex, so one bonnie++ or iozone benchmark won't > > reflect the performance of a other Ceph setup. > > > > Wido > > > > > I really appreciated the help. > > > > > > Thanks in advance, > > > > > > Best regards, > > > > > > *German Anders* > > > > > > > > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > ceph-users mailing list > > > ceph-users@lists.ceph.com (mailto:ceph-users@lists.ceph.com) > > > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > > > > > > > > > -- > > Wido den Hollander > > 42on B.V. > > > > Phone: +31 (0)20 700 9902 (tel:%2B31%20%280%2920%20700%209902) > > Skype: contact42on > > _______________________________________________ > > ceph-users mailing list > > ceph-users@lists.ceph.com (mailto:ceph-users@lists.ceph.com) > > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com (mailto:ceph-users@lists.ceph.com) > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > >
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
