Sorry I didn't get to this very quickly -- in future if you keep your questions on the list, rather than to a single person, you will get faster responses. 2011/1/14 yueluck <yueluck@xxxxxxx>: > because ceph is not stable now, my xen-disk store on ceph. > so i can not use ceph in my product!!! > you do not have much experence here? you are developper,i think? Yes, I am a developer on Ceph. I have limited experience with other storage systems, though, especially those targeted at Xen node storage. Beyond that, there isn't a single "best" solution to this kind of problem. It depends a great deal on your individual needs, and I don't know what they are except that you want to store images. So, I can recommend that you *look at* RBD, which is a part of Ceph but considerably less complicated (and more stable) than the Ceph POSIX-compliant filesystem, and I can recommend that you *look at* Sheepdog, which is designed precisely and only for storing VM images (although I think it's focused on KVM). Determining which of these (or other alternatives that I'm unaware of) is best for you is something you have to do on your own. It will require a lot of work no matter what solution you pick -- if it was an easy problem to solve, somebody would have done so already. -Greg > i am read source codeÂdepend onÂlogs, know outline, but not details. > do youÂread code ? > > ----------------------------------- > AtÂ2011-01-15Â01:18:56ï"GregoryÂFarnum"Â<gregf@xxxxxxxxxxxxxxx>Âwrote: > >>IÂdon'tÂhaveÂmuchÂexperienceÂhere,ÂbutÂifÂCeph'sÂownÂRBDÂdoesn't >>satisfyÂyourÂneedsÂIÂguessÂI'dÂlookÂatÂSheepdog. >> >>2011/1/13ÂyueluckÂ<yueluck@xxxxxxx>: >>>ÂcanÂyouÂRecommendÂmeÂotherÂdfs,likeÂceph.ÂyouÂknowÂorÂhaveÂexperenceÂonÂit. >>>ÂiÂwillÂstoreÂxen-diskÂonÂit.ÂÂÂitÂshouldÂÂstable,multiÂnodes,goodÂThroughput >>> >>> >>> >>>Âthanks >>>Â------------------------------------------------------ >>> >>>ÂAtÂ2011-01-10Â02:05:52ï"GregoryÂFarnum"Â<gregf@xxxxxxxxxxxxxxx>Âwrote: >>> >>>>MovingÂthisÂtoÂtheÂlistÂwhereÂitÂbelongs. >>>> >>>>2011/1/9ÂyueluckÂ<yueluck@xxxxxxx>: >>>>>Â1.ceph-client-standalone.gitÂ,ceph-client.git >>>>>ÂwhatÂisÂtheÂdifferentÂofÂtheÂclient.git???iÂknowÂceph-client-standalone.git >>>>>ÂisÂforÂceph.ko, >>>>>ÂwhatÂceph-client.gitÂdo? >>>>ceph-client.gitÂisÂtheÂbranchÂweÂdoÂallÂourÂin-kernelÂdevelopmentÂon. >>>>ceph-client-standaloneÂisÂtheÂfs/cephÂtreeÂofÂceph-clientÂwith >>>>additionalÂ#ifdefsÂforÂbackporting.ÂIfÂpossibleÂyouÂshouldÂbeÂusing >>>>ceph-clientÂasÂit'sÂmoreÂtestedÂandÂup-to-date. >>>> >>>>>Â2.ÂwhatÂisÂdifferentÂofÂscsi-osd,rados? >>>>>Âscsi-osdÂisÂinÂSCSIÂArchitectureÂModel,ÂtheyÂareÂobject-basedÂstorage?Âif >>>>>ÂradosÂdependÂonÂscsi-osd(osd.ko,libosd.ko.....)? >>>>CephÂOSDsÂareÂObjectÂStorageÂDevices/Daemons.ÂTheyÂareÂconceptually >>>>similarÂtoÂSCSIÂOSDsÂbutÂareÂnotÂatÂallÂtheÂsameÂthing. >>>> >>>>>Â3.doÂyouÂhaveÂcephÂcallÂflow,document(exceptÂwiki),letÂmeÂknowÂceph >>>>>ÂarchitectureÂquickly. >>>>Unfortunately,Âno.ÂHowever,ÂSage'sÂthesisÂ(whichÂisÂavailableÂonÂthe >>>>CephÂwebsite)ÂhasÂaÂprettyÂgoodÂdescriptionÂofÂtheÂprotocols,ÂifÂnot >>>>theÂmoduleÂarchitecture.ÂYouÂcanÂcheckÂthatÂoutÂforÂaÂlotÂmore >>>>informationÂthanÂisÂonÂtheÂwiki.Â:) >>>>-Greg >>> >>> >>> > > > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
