 So, so my name is Billy Field. I'm here to talk to you today about the protection in recovery for OpenStack environments and how to thrive and survive. So we have 15 minutes and I want to cover this quick agenda. I want to kindly direct you to D8. So anything additional beyond this conversation, stop on by. We can dive in deeper. There's only so much we can cover in 15 minute span. I'm going to give you a quick bio about myself to level set to tell you where I'm from. Talk a little bit about the journey to the cloud, right? We're all here at the OpenStack summit, you know, we'll talk a little bit what that means and where things stand. The protection policies and recovery engines, you know, what are the challenges that people are facing in the cloud today? How organizations are attempting to solve these challenges and these problems? How truly it can help solve these problems, right? So this is our driver to the OpenStack community to help the organizations tackle these. A little bit about the architecture of how we're doing what we're doing. And then lastly, what we're going to do is a quick demo. And again, D8, we can get into the nuts and bolts exactly of how we're doing it, okay? So 20 seconds about myself just to get a baseline. So I run the cloud architecture team at Trilio Data. Quick little activity at EMC for 12 plus years. So I've been focused in the data protection and cloud architecture space for the last decade or so. And where the journey always starts here is, I've talked to several dozens of customers and prospects over at our booth and there's a continuum of the journey to the cloud, right? I've heard customers say, hey, I'm just doing discovery today. I'm still doing some research on where things stand if I'm going to go from VMware over to OpenStack. I've talked to people that say, hey, we've limited it down to a couple different distributions and I'm actively doing some POCs, right? We have some customers and prospects that are like, hey, I'm in non-prod, not quite in production, but I'm almost there. And we also have people that have been run to OpenStack for the last three plus years, right? So no matter where you are on this continuum, the business is going to drive the assurance of it. Is my data protected? Can I get back to any point in time within that cloud with some sort of a solution? So instead of killing you by death by a power point with bullet after bullet after bullet, let's kind of settle it down here. Let me introduce you to Tom, right? Tom's the director of IT at Acmecorp and he sees all cloud operations. He's gone and convinced the business. He's built this great OpenStack cloud. It has all the cloud attributes, right? It scales. It's multi-tenant. It's self-service. It's great. The business loves the flexibility. The business is really kind of liking how they have a good TCO ROI on the solution. And overall, Tom, you know, he's done his due diligence, but his life is pretty darn good. He's like, I've got the feather in my cap. My cloud's up and running. The business likes it. I'm good. So let's talk a little bit about the challenges in Tom's near-term future here. As you can see, Tom's mood mood here is he's a happy guy. Okay. Enter Sam, right? Sam's in DevOps and Sam comes to Tom and says, hey, nice shiny new cloud. Talk to me a little bit about testing and migration. I've got a lot of stuff that I need to do. I got iterations of things I need to spin up and spin down. And Tom says, hadn't quite thought about that. Interesting conversation, right? We'll table that for a second and get back to it. Next, you know, Tony comes by. Tony's from the app team, right? We all know those to app guys, right? Mission critical, building the next gen applications, you know, thriving the pocketbook for the business, you know, next release, next release, next release. And hey, Tony says to Tom, all my stuff's in your cloud. What about backup and recovery? How do I do that? Tom's mood meter. All right, hadn't quite gotten there. Not quite sure about that. Let me table that, too, because I haven't gotten there yet. And Eric, the corporate guys, right? We all know what these guys do, right? Audits, SLAs, RPOs, RTOs. So Tom's having a really bad day because he's like, I'm in trouble. This cloud can't address any of my team's needs, right? So Tom reaches into his tool belt and he's got a couple different tools here. I can go ahead, go the scripting approach, so I can go to each business unit and start developing scripts and go ahead and compile all these things and piece them together. And I got a new slide. Give me a second here. So a little snafu. I had to switch laptops here, but long story short. So Tom has scripting available on his tool belt. He's got schedule scripting, which is crontabs. He modifies cron jobs across all these different machines. I'm sorry. He can leverage any sort of array-based technology, some volume-level snapshots. He can dump to NFS if he wants to. He can use host-based tools. He could use legacy approach, such as backup, backup to disk, backup to tape, you name it. Or he can just go straight application tool sets, right? But the dirty little secret is none of these approach are built with a cloud in mind, right? They don't have any cloud characteristics, and much like the duct tape, you know, you're stitching these things together, like bubblegum and shoelaces, right? So the challenge is if he deploys this, he'll run into a couple different things. Number one, he's going to be managing a bunch of different backup sets. He's going to be going through managing Excel documents and where things live, and really a management nightmare here. He's also got to basically maintain every single point in time for the VM configurations in the sense of the flavor, the volume types, you name it, the compute resources. In addition to that, what he has to do is he has to go through and then map every single volume associated with that point in time to those VMs, which could be hundreds, if not thousands, right? So this is kind of a snowball effect. He's getting to the point where not quite sure, and then on top of it, he's got all the security groups and the network policies and the net mask and IP addresses, right? So if he goes with the legacy-based approach, these are his challenges, right? We've all been here, right? We can all agree that we've all done this in the infrastructure. So let me introduce you to Trilio, okay? Trilio is the only native data protection solution for OpenStack, and what that means is we are built from the ground up with the cloud in mind, and to the right hand side are all our attributes, that every cloud is to be deployed and not only to have protection for us. So we are downloadable, we're agentless, multi-tenant, inscalable, flexible, non-disruptive, incremental capture forever, and then the last bullet point is key, OpenStack native. So what do we do? What do we capture, okay? Do we capture just the volume? No. Do we capture just the VM? No. What we do is we capture a couple different components. We get the exact OpenStack configuration metadata as a first step, and what that means is all the neutron components, so the net mask, the firewall settings, the IP addresses, you name it, okay? We get the NOVA components, anything to do with the compute layer. We get all the cinder configurations, so we get the volume types, the volume configurations, and lastly we grab the image from glance. So think about that as the configuration, and then on top of that what happens is we go through, we then grab the actual data from the OpenStack environment as a phase two. So what we do here at Trilio is we do a full once in the incremental forever capture, and what we do is we take our baseline level zero, okay? We copy it to a target repository, which is NFS, Swift, and we've recently announced Ceph Gateway and Amazon S3 for this quarter, and what we do is we get our base level zero, and then every subsequent job is incremental forever. Now the best part about our policies is all driven by the tenant, so the whole self-serviceability component of the cloud, we port over to the data protection realm, right? So me as a tenant, I spin up my infrastructure, I spin up my applications, and then now I can have a new tab in Horizon that basically defines my service level agreements for my workloads, okay? So this is just a little infographic that says we do a full once, we leverage snapshot based technology, we then do incremental forever at the block level capture, and then what happens when we get the data under management or backed up? I'm sorry, this is a new deck here, it's spinning through on me, but backup data protection is only one component of the engine, right? Recovery is the other, right? I've got to be able to recover that data. So we offer four different types of recovery today. The first as you can see is in place recovery. Now what that means is I go to my I go to my cloud, I log in, the instances are either corrupt or terminated, they're gone, right? So I can go to Trilio, I go to my workload, I go to my backups, and I say in place recovery, and what we do is we recover the metadata and the data to that point in time, in place as to where it was captured from, okay? The next, again, so once we get that data, metadata and data, now we have the ability to do a selective restore, right? So I can go ahead and say, hey, maybe I want to spin this data set up to a different availability zone or a different region altogether, maybe with just different parameters, okay? Sorry, I'm kind of flying through here because it's on loop for some reason and I can't change it. So once you get the recovery into a component of a different availability zone, a different region, different configuration, you start talking a little bit about disaster recovery and migration scenarios, right? So I can spin that environment up into a new cloud, right? So I can go from, you know, my private cloud here at site A, and I can spin it up over here into site B, whether it's private, public, or hybrid, you pick or choose, right? And then the last use case with us is we talk about VM level, VMVM block block. We also expose the ability to recover files or folder level recovery, again, from the tenant. So I can mount the data set up. I can browse files, directories, or folders. I can use SCP or SSHFS and get the files or folders directly to a local workstation or back into my source location. So Trilio architecture, what does it look like? How do we do what we do? So to the top left-hand corner, what you'll see is we will ship our virtual appliance in native QCOW 2 format, and you will deploy that on a standalone KVM box. Once that is deployed, what we do is on each one of the controllers, we will install our service endpoint, our Trilio endpoint. And then the last component is our lightweight Python-based data mover, which sits on top of the compute nodes. So the good thing about this component is that I can start today with a 10-node open-stat cluster with 10 data movers, and then I have the ability to scale. That's the point of these clouds, right? I can start here in a 10-node, scale to 100, if not more, and then just enable this data mover on each one of the compute nodes. Now, the nice part about our solution is you can deploy this as a standalone. You can just dot slash, run this, run it once, but we also provide Ansible scripts as well. So if you want to pump this into Ansible to automatically scale as a DevOps approach to scale this cloud out, we can go in with that approach as well. So again, we only had 15, 20 minutes here. So in summary, everyone here is on the cloud journey to the cloud, right? Whether it's in its infancy or you've been straight running OpenStack for the last few years, your business units are demanding data protection. This is no longer if it works, it's when it works, and this is more of a compliance and security measure that the OpenStack community needs to adopt to. Trilio is the native OpenStack data protection tool that will empower this business challenge. And the key features, the cloud agnostic features, is that we are agentless, we're incremental forever capture, we're tenant-driven, and again, we scale much like your cloud should scale, right? And last but not least is we do solve that challenge. We do solve that data protection challenge that you do have for OpenStack. Okay, quick, if you're joining this session, there's three here that I want to kind of clue you in on. One happened yesterday, Ganshem at NEC did a great talk on the strategies of backup OpenStack resources. In a couple hours, Murley, our CTO, is going to be tapping the strengths of hybrid cloud, and then tomorrow he's going to be doing a follow-up session for a sneak peek for that hybrid cloud offering. So open up your app, add those if you're interested, but I think you'll definitely find some good value here. Okay, we'll jump into a demo here. Okay, so before I play this, I recorded this because I was unsure of network connectivity, and I'm not a big fan of dropping connections via demo here. The first thing I really want I want to cover three things. Number one is how simple it is to manage our product, quickly go over how it is to protect, and lastly how to recover because that's really, you know, the core functionality of Trilio. And setting the table, what you'll see here is we're directly tied into Horizon. So we provide a Horizon plugin, and we're compatible across all distributions of OpenStack. Yes, we do have a REST API. Yes, we do have a full CLI, but for visual purposes here, I'll show you our integration here with Horizon. So I'm Tom, right? So I'm logged in as Acme Co-Tenant, and I'm logged in as Tom. And you see that backups tab here that we've exposed at the top panel. Now what I've done here is I have my service level agreements that I've defined for my cloud. So you can see gold, silver, bronze, you can kind of tell based on those policies what they do. So the gold policy here has a 3VM workload. Looking at the snapshots tab, there's your full, and then we have two incrementals. Again, so we're incremental forever, as you can see the size of those. If we dive into each one of those snapshots, what you do is you'll see the metadata associated with that. So we get the security groups, the flavors, the volumes, the networks. So we get the exact point in time configuration of that capture job, in addition to the actual data itself. We'll show you the status of the job, excuse me, the status of the job, how long it took, the size. And basically what you can do is go SLA to SLA, and the good thing is the tenants here can define their own workloads. I just showed a couple just for demo purposes. On the flip side, so when you get the data protected, you've got a couple different things. There's the one-click restore, which is an in-place recovery. Additionally, we provide this functionality called a selective restore. So again, this is a point in time recovery that I can go ahead. The neat thing here is I can go and define a different availability zone. So I captured from an east coast availability zone, now I can spin it up to a different availability zone. Maybe I captured it here, network wise on a 10.network, I can spin it up and put on an 11.network. The capture of the three VM instances are listed there underneath the instances. The cool thing about this is I can go ahead and rename the instances, but also give it a different flavor. So if it's a DR repository or another region that has less compute or less resources, I can say, hey, this is an extra large resource, let me drop it to a medium or let me drop it to a tiny. And then last but not least, you can go through and say, hey, maybe I want to just restore all the VMs or just one VM. You can pick or choose. And then what happens is we click this banner and we start the recovery process after we hit submit. Now going back to this whole VM capture, block level capture, the last thing we do here is you'll see the mount snapshot operation. What we do is we'll ship out a file recovery manager that what you'll do is when you mount, you're going to logically take the volume structure of that point in time. You're going to logically map it to that file manager. It'll scan for the new devices. And then what you can do is you can hit the IP address of that file manager, web browse the volume structures. You can download them locally. You can use SSHFS or SCP to grab that data and so on and so forth. Fast forwarded here. We talked about tenant, tenant, tenant. Here's the admin view. So I can go here globally, look at all my workloads. I can take a look at how many VMs. I can look at performance metrics. I can look at where these backup jobs and data protection jobs are writing to. Underneath the nodes tab, these are our virtual appliances. We list all our data movers. Again, these are the components that go under the compute resources. So in this case, I have seven computer data movers. We list all the storage repositories. Stop there. Sorry about that. Sorry, David. Delete my CEO's emails. I'm sure he's not happy about that. And then last but not least, you've got audit logs. You've got system configurations. You've got settings and you've got licensing underneath admin panel. So again, can't really give a detailed demo in five minutes, but please stop by D8. We can go into the nuts and bolts and dive deeper technically as to what Trilio data is doing. So thank you very much.