[ceph-users] capacity planing with SSD Cache Pool Tiering
Hi folks, one more question: after some more interanal discussions, I'm faced with the question how a SSD Cache Pool Tiering is calculated in the "overall" usable storage space. And how "big" do I calculate an SSD Cache Pool? From my understanding, the cache pool is not calculated into the overall usable space. It is a "cache". E.g. The slow pool is 100 TB, the SSD Cache 10 TB, I dont have 110TB all in all? True? I'm wrong? As always thanks a lot and regards! Götz -- Götz Reinicke IT-Koordinator Tel. +49 7141 969 82 420 E-Mail goetz.reini...@filmakademie.de Filmakademie Baden-Württemberg GmbH Akademiehof 10 71638 Ludwigsburg www.filmakademie.de Eintragung Amtsgericht Stuttgart HRB 205016 Vorsitzender des Aufsichtsrats: Jürgen Walter MdL Staatssekretär im Ministerium für Wissenschaft, Forschung und Kunst Baden-Württemberg Geschäftsführer: Prof. Thomas Schadt smime.p7s Description: S/MIME Cryptographic Signature ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] capacity planing with SSD Cache Pool Tiering
Hi, The cache doesn't give you any additional storage capacity as the cache can never store data, thats not on the tier below it (or store more writes than the underlying storage has room for). As for how much you should go for... thats very much up to your use case. Try to come up with an estimate of how much data is frequently being accessed (this is the data most likely to remain in the cache). Then double that estimate - ALWAYS double your estimates ;) (this isn't Ceph-specific). There might be additional magic, but in general the cache will store all the data that is being read from the underlying storage (in hopes of it being required again later) as well as any writes that may occur (if you don't configure the cache to be read-only that is). Do note that this also means that currently (afaik this is being worked on) pulling a backup of your RBDs will completely flush the cache. This only means that the files you'd want cached will have to be pulled back in after that and you may lose the performance advantage for a little while after each backup. Hope that helps, dont hesitate with further inquiries! Marc On 05/05/2015 11:05 AM, Götz Reinicke - IT Koordinator wrote: Hi folks, one more question: after some more interanal discussions, I'm faced with the question how a SSD Cache Pool Tiering is calculated in the "overall" usable storage space. And how "big" do I calculate an SSD Cache Pool? From my understanding, the cache pool is not calculated into the overall usable space. It is a "cache". E.g. The slow pool is 100 TB, the SSD Cache 10 TB, I dont have 110TB all in all? True? I'm wrong? As always thanks a lot and regards! Götz ___ ceph-users mailing list 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
Re: [ceph-users] capacity planing with SSD Cache Pool Tiering
Thanks Marc & Nick, that makes things much more clear! /Götz Am 05.05.15 um 11:36 schrieb Nick Fisk: > Just to add, the caching promote/demotes whole objects, so if you have > lots of small random IO’s you will need a lot more cache than compared > to the actual amount of hot data. Reducing the RBD object size can help > with this, but YMMV > > > > Also don’t try and compare Ceph Tiering to a generic cache. With a > generic cache you tend to get a benefit even when the cache is too > small, however due to the way Ceph promote/demotes, cache misses are > very expensive and I have found that unless the bulk of you’re working > set fits in the cache tier, then performance can actually be worse than > without the cache. > > > > *From:*ceph-users [mailto:ceph-users-boun...@lists.ceph.com] *On Behalf > Of *Marc > *Sent:* 05 May 2015 10:25 > *To:* ceph-users@lists.ceph.com; goetz.reini...@filmakademie.de > *Subject:* Re: [ceph-users] capacity planing with SSD Cache Pool Tiering > > > > Hi, > > The cache doesn't give you any additional storage capacity as the cache > can never store data, thats not on the tier below it (or store more > writes than the underlying storage has room for). > > As for how much you should go for... thats very much up to your use > case. Try to come up with an estimate of how much data is frequently > being accessed (this is the data most likely to remain in the cache). > Then double that estimate - ALWAYS double your estimates ;) (this isn't > Ceph-specific). > > There might be additional magic, but in general the cache will store all > the data that is being read from the underlying storage (in hopes of it > being required again later) as well as any writes that may occur (if you > don't configure the cache to be read-only that is). Do note that this > also means that currently (afaik this is being worked on) pulling a > backup of your RBDs will completely flush the cache. This only means > that the files you'd want cached will have to be pulled back in after > that and you may lose the performance advantage for a little while after > each backup. > > Hope that helps, dont hesitate with further inquiries! > > > Marc -- Götz Reinicke IT-Koordinator Tel. +49 7141 969 82 420 E-Mail goetz.reini...@filmakademie.de Filmakademie Baden-Württemberg GmbH Akademiehof 10 71638 Ludwigsburg www.filmakademie.de Eintragung Amtsgericht Stuttgart HRB 205016 Vorsitzender des Aufsichtsrats: Jürgen Walter MdL Staatssekretär im Ministerium für Wissenschaft, Forschung und Kunst Baden-Württemberg Geschäftsführer: Prof. Thomas Schadt smime.p7s Description: S/MIME Cryptographic Signature ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
