Brett gives you some reasons why he thinks CephFS is the best distributed file system.
So there's a couple things that make CephFS the best distributed file system, but first of all it is fully POSIX compliant and it's got full support for Linux extended attributes. So this allows for endless integration into existing Linux applications.
So if you've got Linux clients, don't worry, mount this like it's a local filesystem and it's gonna work great (obviously its a network file system but you see my point). And second of all, what's really fun is it integrates with Samba very well which means flawless access to Windows clients into the cluster. Not only can you have windows access, you can have full windows access control list control over the permissions which is a big draw for a lot of end users.
So the next great thing in CephFS is its snapshots. So just about every distributed file system now does snapshots. What makes Ceph’s so great are that they're quick and easy with emphasis on the quick. I don't know if you've ever used GlusterFS-- it takes a minute for it to snap everything and close it off. Ceph will complete its snapshots *snaps finger* like that, obviously as your file system grows it takes a little longer, but it's very very quick.
Second of all is how easy it is to just get to get to your snapshot unlike some other applications, where you'd have to clone it and mount it or something like that. It is the path of what your snapshotting and then just a PennDOT snap to the end of it in the name of snapshot, and there's all your files. So, while that's really easy, what really adds and makes it greater is its integration into Samba as I mentioned earlier, plays very nice with Shadow Copy, so Shadow Copy is version control. Users can just access their own snapshots and get their old files back, so it just really lightens the load on the admin team, should you have to snapshot a bunch of stuff and retrieve files that people have deleted or lost.
Alright, so the last thing I want to talk about that makes CephFS great is its ability to do directory pinning. So to understand that, first you have to understand the way Ceph organizes its data is into storage pools. Storage pools are separated by their data protection type, whether that's replication or a erasure coding, or by the device that it's on, whether it's hard drives or SSD’s.
So what that means is if CephFS can use multiple storage pools under one file system namespace, you can have hard drives and SSD's in the same file system, or you can have scratch storage with one replica, and then another pool with a really big nice erasure coding all in one file system, and you can pin directories in your file system to live on certain pools. So, that means say you're an institution that needs a little bit of really fast storage for maybe their video editing team, but then everything else is just kind of gonna be set-it, forget-it type of data, you could have a big tough cluster, a little bit of SSD storage; they would see one file system, one share, but one directory would be fast for the editors, and everything else would be slower, more economical eraser coated hard drives.
Okay, so that's some of the best parts of CephFS. Fun-fact about Ceph, some people wonder what the name is, “C E P H”, it's short for cephalopod, which is the family class (I don't know I'm not a marine biologist) for octopuses and squid, those fun little multi legged creatures that swim in the ocean. But the point is they have many legs, like Ceph it has many pieces, it loses one, it still rebuilds and moves on.
So if you want to learn more about CephFS or Ceph in general, check out some of our other videos that we've done discussing the topics. Check out our knowledgebase; we have some of our experiences written up there to help. Of course, the Ceph official documentation is solid, as well as the community as a whole, the mailing list, and just everything you can find on the Internet. I find it's one of the strongest open-source communities ongoing.
Alright, well thanks for watching another weekly tech tip. Throw any comments, questions, or anything below in the comments, or email us at info@45drives, or Instagram, or Twitter (whatever social media we're on). If you've got any ideas of any other tips that you want to see personally, we'd love to do it for you. Thanks for watching.
Contact us to discuss your storage needs and to find out why the Storinator is right for your business.Contact 45Drives