> On 05 Sep 2014, at 10:30, Nigel Williams <nigel.d.williams at gmail.com> wrote: > > On Fri, Sep 5, 2014 at 5:46 PM, Dan Van Der Ster > <daniel.vanderster at cern.ch> wrote: >>> On 05 Sep 2014, at 03:09, Christian Balzer <chibi at gol.com> wrote: >>> You might want to look into cache pools (and dedicated SSD servers with >>> fast controllers and CPUs) in your test cluster and for the future. >>> Right now my impression is that there is quite a bit more polishing to be >>> done (retention of hot objects, etc) and there have been stability concerns >>> raised here. >> >> Right, Greg already said publicly not to use the cache tiers for RBD. > > I lost the context for this statement you reference from Greg > (presumably Greg Farnum?) - was it a reference to bcache or Ceph cache > tiering? Could you point me to where it was stated please. Cache tiering. I was referring to this thread back around when firefly was released. http://lists.ceph.com/pipermail/ceph-users-ceph.com/2014-May/039504.html > At present, the cache pools are fairly limited in their real-world usefulness. > ... > 3) The cost of a cache miss is pretty high, so they should only be > used when the active set fits within the cache and doesn't change too > frequently. > ... in general, > I would only explore cache pools if you expect to periodically pull in > working data sets out of much larger sets of cold data (e.g., jobs run > against a particular bit of scientific data out of your entire > archive). There should be a wealth of real world experience with RBD and cache tiering by now, and I admit that I haven?t followed that line of development. Is anyone running RBD through a cache tier and getting good performance with it? Cheers, Dan
 |