The first episode of Raising the Floor, the enterprise datacenter podcast from Foskett Services, features thought leaders from two key cloud storage-enabling companies:
We continue our conversation with a discussion of the mechanics of cloud storage. How can real-world applications make use of storage connected over the Internet?
Stephen Foskett: That’s one thing that’s interesting is that the cloud storage is not just conventional storage. There’s actually a very different architecture behind it as you mentioned in terms of how it’s actually constructed. There’s also the fact that the access method is very different. That maybe something we should talk about as well. Andres, do you want to take a moment to talk about object storage and REST?
Andres Rodriguez: Absolutely, and I should say before Nasuni, I actually did a company called Archivas to develop an object store. What we did there is the concept is rather than having block level read/write fast access, you develop a system that’s going to have blob level. That is much larger binaries, typically, at least a few hundred kilobytes in size. The binaries are accessed into the cluster, into the object store in a transactional way. Just like HTTP, in a file transfer way. What you have is this system that exists out there that you can access through essentially HTTP, get, could, and delete which are very gross operations, very high-level operations.
You’re going to be putting out all these binaries into those systems. What is important to understand those systems, not only do you have the latency’s built into protocols like that that are really designed to be accessed to be used over the Wide Area Network.
But, the semantics at the other end are such that those are protocols that are very good if you want to create a blob, get a blob back, or delete it, but not if you want to modify it.
Now, if we’re talking in terms of trying to make that compatible with file systems, there’s a real leap that you have to take so that you can have something that is read/write very fast coexist with that kind of backend.
Stephen: Yes, and I think that that’s really the next point we need to make. Most conventional applications are not compatible with this HTTP REST interface that most public cloud providers rely on. Now, there are more applications all the time that are compatible with it but does this mean that cloud is off-limits that you can’t use it for conventional applications?
Josh Goldstein: Not at all. I think that the thing is you need a technology and this is exactly that the business that Nasuni sort have gone into. You need a technology to pry in the cloud to make it appear the same way that you would expect your local storage systems to appear. And that is what’s really critical is laying this on top of the object store so that you get the best of both worlds. You are now getting the scalability and the pay as you go pricing that you can get from a number of different public cloud providers but at the same time you are in and out reading with that cloud in a way that is very familiar to you and compatible with your applications.
Stephen: I think that is one of the interesting things about these two companies. And one of the reasons that I picked you guys to talk together is because I see you guys as basically two sides of a coin. You have block storage and you have files storage. Do you want to talk a little bit each of you about what it is exactly that you are doing with public clouds storage?
Andres: Sure thing. So, I’ll start there and I said anything that traditional storage world, we have file based systems. We have block based systems and that is exactly where Cirtas and Nasuni stand. We are the equivalent of say a company named EMC and NetApp for the cloud world. The approaches are complimentary. And they are both trying to solve the same problem. I’ll start on the file side but Josh can take on the block. But on the file level is really, look, if you want to have something that behaves very much like a file server, say like a NetApp box. It means you are going to have a file system, you want to have a protocol to export it locally on so something like CIFS, you are now going to have access to directory integration so that you can have access control. This is what makes a file server useful in the datacenter.
And furthermore, remember what I said about no backups? If you are not… If the cloud is completely reliable right. Then what you need to have on top of the cloud is some kind of system that allows you to version all the changes out to the cloud. And you want to do that for several reasons.
But one of the most important reasons is so you can go back to your file system as it was at any point in time or any specific file or directory at any point of time. And we’ll store it without having to go to back up by just relying on this device. In our case, on our Nasuni file server and the cloud back-end.
In addition to that and I let Josh kind of paint the full picture from the block perspective.
You also want to use this kind of snapshotting technology to make sure that you are only sending to the cloud and storing in the cloud because you are now going to be paying for this for the changes or the minimum amount of information but you can’t. And so you want to do a very good job identifying any kind of duplication of the data.
And then de-duplicate that data, compressing that data and incorporating all of that data before it lands in the cloud. We can talk afterwards about security and encryption but that’s really what makes the bridge. I’m skipping the file system side.