What is Multi-Cloud Object Storage?

- Hi, I'm Mike Matchett with Small World Big Data. And we're gonna talk about object storage today. But not just little object storage, object storage is never little, we're gonna talk about big object storage. In fact object storage that spans cloud. We're gonna talk about multi-cloud object storage which is where the world's heading. As our data gets bigger and bigger and bigger, as everyone gets more and more data, we're gonna have to store it in a number of places all at the same time but that is gonna give us a whole lot of challenges. One, we're gonna have multiple clouds, we're gonna have on-premise, we're gonna have, we're gonna have private cloud, we're gonna have us hybrid storage nightmare of where is the data, where is the master copy, where do I put my application at, and how can I get at it and then gonna get into all sorts of things and data validation, data protection, and that. So world's getting messier however, we've got someone here today. He's gonna tell us how they're helping us solve that problem. I've got Joe Arnold with SwiftStack and he's got some new announcements telling us how we're gonna manage multi-cloud. Welcome, Joe.

- Hey, thanks Mike.

What is the SwiftStack Object Storage System?

- Alright, so for the people who aren't totally familiar with SwiftStack, and the Swift Object Storage system, what are you guys doing? What are you bringing to market with Swift in SwiftStack?

- Yeah, that's a good question. So what we have, we have a software to find storage product and really, what we're doing now is multi-cloud data management. So we have this object storage system that supports both file and object. And we introduced the file system about six months ago so people can have file and object for the same data. We're really focused on big data footprints and we try to make it as easy to deploy, operate scale, all that, the stuff so people can run that in their private data centers. And looking forward to talking about the multi-cloud stuff that we're, that we've been doing and we're introducing some new features around that.

- Alright, so, so if I were to talk to you a year or two ago, we'd be talking about putting Swift as an object storage or on my premise so that I could almost, as an enterprise shop, compete with Amazon S3 or compete with a cloud provider's object storage and say, to avoid that shadow IT situation, now it's probably pretty common couple of years ago. But today we start acknowledging, hey, no, you're gonna use both, you got to. There's bursting, there's compute services in the cloud. You were telling me that lots of people wanna do cloud analytics. There's services like transliteration stuff, you just gotta have your data there and here. So you can't, it's not us versus them, it's how do I marry those things together. And I think that's the Genesis now that you're gonna tell us about for OneSpace. So what does OneSpace do for us?

- Yeah, we're actually, you know, customers, what we really wanted to do, just more with their data. They wanted to do analytics on it, they wanted to distribute it. And look, they don't have necessarily the compute capacity on-premises that is available on the public cloud. And there's all sorts of really great services in the public cloud that they wanna leverage but for analytics and machine learning and AI, I know those are kind of buzzwords but there really are some compelling services up there. So how do you get that data there? And we're introducing a product called OneSpace. It's on open source project too, just like the core of our storage, Swift. And what it does is it creates a single namespace between two clouds. And when you have that namespace between clouds, that means you can access the data either on-premises or in the public cloud and you can do all sorts of really interesting things. Like you can burst applications out to the public cloud, you can do live migrations, you can live cycle data and make that just a transparent experience for applications. And that gives tons of flexibility for applications and these large data use cases to then go on and consume the services in the public cloud.

Object storage on-premise and in the cloud

- So you become with one, you actually become the, this data storage, object storage service layer whether my app is running on-premise or running in a cloud, and then you're gonna behind-the-scenes marry the storage of data that I might have in the cloud with the storage of data I have on-premise and make that seamless, right?

- Yeah.

- So it's kind of, and. Alright, so that sounds easy enough. Can't be that easy, right? Because you've mentioned live migration, and you've mentioned couple other things, and might get easy to toss those words out but alright let's step back because, that, those aren't easy things. Live migrations itself it's gotta have many different problems to it, right?

- Right, 'cause you have to coordinate, okay, I have an application and you have these large data sets. We're beyond the days of when we can take a Sunday afternoon and tell people to stop using the service. Now that's not gonna work when it's a multi-petabyte or hundreds of terabytes of data. It takes weeks and months to move that data. So how do you create a framework so that data can, applications can move from one site to another while still being able to access the data. So that means you have to take in account things like, okay what happens when data is deleted, how do you, when data is updated, which one's the latest, and we have to coordinate all that. So fortunately, we've had this experience to deal with this issue with our distributed storage system across multiple sites. And that's one of the things about the SwiftStack storage system is you can do multi-site. And we're adding a site, adding multi-cloud to this. And we're using that same logic to keep all that data in sync. And yeah, there's a lot of engineering that goes into it. We have some really large web scale customers that are using it right now. We have financial services that are interested in moving that data out to the public cloud and keeping all that data in sync. It's a real challenge to pull off and, but that's what we've done.

- You guys, you guys don't have large semi-trailers that you're, shipping hard disks back and forth with, right?

- We actually, it was really funny, we had a customer conversation where they wanted to do something really particular, and I'm like, you realize that API charges for what you wanna do is gonna cost $50000 and the data storage is gonna cost $50, right, for this one particular use case. So you really kinda have to know how to use each one of these clouds in a particular way and know how to store data into it. When the life cycle data off, how to control the access to it. And there's actually even some, almost pricing hacks, if you will, that you can start to execute. So you think about doing a computeverse workload. You might have a large footprint of data and well why move data unless you have to, right? You, you can burst up the compute workloads, get that single namespace of the data and only pull off data as you need it. And then once you're done, don't ship all that data out, just drop it on the floor, get the answer to what you need, and then ship that back. So you're not necessarily having to pay the egress charges for all the data.

- Yeah because in a way it becomes the, where the data sits becomes demand-driven on the usage of it, under the scenes rather than trying to have to plan or predict or make multiple copies of it. 'Cause you don't really wanna be moving that petabyte around unless you have to, right?

Moving data between clouds

- Yeah, and it's also interesting, even on reverse, right? So people wanna migrate, say they have an application team, they build an application, they're running it in Android OS. But then for operational reasons they wanna migrate that application back, well that happens all the time, right? Think about industries where some, like hit-driven, like gaming for example where they might now know the how much capacity they're gonna need to run that service. So yeah, deployed in the public cloud, get it running and then when they're ready to brick that back on site, well, now we can create that namespace, they can move the data as part of that application back, and it's totally seamless. There's no interruption to the service of the application.

- Alright that was the big thing, you're not, you're not, you're not, you do not send these cut points and doing replication points, and doing all this other stuff that has to be managed. It's just happening under the scene, which is so cool. Alright, there's a ton more of this. I know you told me you had like, you had hours of video and white papers to get more into that. We only got another maybe 30 seconds here but I wanted, you also told me something about metadata component that you also have there at SwiftStack. What's going on with metadata these days?

- Yeah, I mean, when you start having so much data, organizing it with volumes and folders really starts to break down. So we've introduced a way to index all the metadata that is associated with the objects. And then once that data has been indexed and searchable, now we have, we have a SwiftStack client which is a desktop client, runs on Windows, Mac and Linux environments. And you can do metadata searching on it. So you have data all across your environment, type in the keywords you wanna find, or build up sophisticated queries and that you can find your data.

- Yeah we know that you can probably come up with use cases in genetics and financial services where they have all sorts of hairy metadata and everything. But I'm just thinking of the poor IT guy who just happens to have a million back up files stored as object, somewhere like S3, and he's like I gotta restore or find one of them and there's no way to get to it. So this could, this would solve their problem 'cause now you can just type it and say I want this back up for this machine and get to where those are. Not to to get into the data protection that you're gonna make obsolete anyway, the backups you're gonna make obsolete, with what you're doing but. So thanks, thanks, thanks so much. Obviously SwiftStack, is there a place where you wanna come, come to your website and get more information? Is that the right thing to do or just download Swift and start at it?

- Well, I mean, I would, check out our YouTube channel. Go to youtube.com/swiftstack and you can take a look at some of the videos and demonstration of what we have. Of course go to the website, we have swiftstack.com/testdrive. You can go, people can try SwiftStack without talking to a salesperson. You can just go to the sign up, try it. We have our product that part of it runs in the cloud and part of it you can download to, to run on some servers to try it out for yourselves. I encourage folks to do that.

- Awesome! I can't wait to do it myself. Thank you for being on the show today, Joe.

- Alright, appreciate it. Thanks, Mike.

- Alright, and thank you. Again, this is Mike Matchett from Small World Big Data. And I'm sure we're gonna hear more about SwiftStack as they come up with some cool new stuff. They've got a huge roadmap they were telling me about. And we'll talk to you guys later, bye.