Hi, Currently, our ceph cluster is all 3-way replicated, and works very nicely. We're consider the possibility of adding an erasure-coding pool; which I understand would require a cache tier in front of it to ensure decent performance. I am wondering what sort of spec should we be thinking about for the cache tier, and how robust erasure-coding pools are compared with the simpler replicated pools? Our test cluster has 3 storage nodes, each with 60 x 6TB RRD, with NVME for journals. So for cache tier, we should presumably be thinking SSD? Roughly how much, and should we consider NVME, or is it not worth the $$$$? Thanks :) Matthew -- The Wellcome Trust Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE. _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |