http://wiki.define-technology.com/mediawiki-1.35.0/index.php?action=history&feed=atom&title=Ceph%3ABest_practices 2026-05-04T20:20:29Z Revision history for this page on the wiki MediaWiki 1.35.0 http://wiki.define-technology.com/mediawiki-1.35.0/index.php?title=Ceph:Best_practices&diff=5016&oldid=prev 2014-10-22T09:47:46Z

<p>Created page with &quot;== Best Practices == Plan it!!! Requirements: size, performance, budget Cost - vs reliability node failures Rack failure? Multi site replication Plan for redundancy of monit...&quot;</p> <p><b>New page</b></p><div>== Best Practices ==<br /> <br /> Plan it!!!<br /> Requirements: size, performance, budget<br /> <br /> Cost - vs reliability<br /> node failures<br /> Rack failure?<br /> Multi site replication<br /> Plan for redundancy of monitors nodes<br /> 3 copies = 8 nides of availablity = less than 1 second of downtime per year....<br /> Everything effects performance - in above and below ceph<br /> Understand your workload<br /> <br /> NUMA and NUNA can effect performance<br /> Multi site - can cause latency issues, applications could time out...<br /> <br /> <br /> == Recommended Storage server config ==<br /> <br /> CPU - 1 core GHz per OSD<br /> Memory - 2GB per OSD<br /> Disk Controller - no extender for data and journal, RAID1 for OS disks<br /> Data - size doesn&#039;t matter but 12 disks per OSD is usually optimum<br /> Journal Disks - SSDs, write optimized<br /> <br /> == NUMA ==<br /> Populate all channels to maximise bandwidth<br /> Use sensible memory configs</div>

Michael