 Lung addiction just doesn't affect you but others around you. Hi Hi, Benjamin Tell us about the last time you tried to communicate with others. It was bad, Carl No one understood me It was like I was speaking a completely different language Hey, Ben We need two terabytes to spin up a thousand VMs for a test rewriting tomorrow two terabytes a thousand VMs today Are you out of your mind? I will be manually tuning storage and reshuffling VMs all night again Two terabytes. Okay. I can do that. Uh tier zero tier one raid. What raid level are you looking for? I don't know man. I just need two terabytes reads or writes sequential to random. What about stripe width snap out of it I just need two terabytes Benjamin come back to me, buddy. I'm right here I'm right here and go That was kind of my wake-up call, you know When I realized I was stuck And now that you have tindry well everything is so much simpler. I'm thinking more clearly and Communicating with my co-workers. I have friends Ben If I were to ask you to take two terabytes and spin them up to a thousand VMs, what would you say now? I'd say Consider it done, dude. Healthier storage Healthier relationships I'm proud of you, buddy Welcome keep storage simple Good afternoon everybody my name is Chuck and I used to have a problem with Luns and I know I'm not alone Because with traditional storage Even if you don't know what a LUN or volume is at the physical level It's causing problems in your cloud because it's providing performance problems Things you're not even aware of these are the unknown unknowns That are causing problems in your environment and I'm going to talk to you a little bit about Tintry what we do and then I'm going to show I'm going to hand it over to Chris who's going to show you a demo of our integration with with Cinder but first Let's talk about Tintry We sell Storage appliances into enterprises service providers Public cloud and when I meet mean by appliance is a physical storage array. So these are all for you boxes with SSDs and HDDs and a hybrid architecture But our unique value is that we are VM aware storage. We actually don't support physical workloads We're not designed for Being a filer what we're designed for is to accelerate and provide performance to virtual machines and we do that We started selling the VM store in 2011 as a solution for vSphere Virtualized environments, we've since added support for Hyper-V Red Hat Enterprise virtualization, which is Red Hat's KVM Virtualization infrastructure and today at the show and at our booth later on and on stage in a few moments We're going to show you our integrations with OpenStack Cinder We get a lot of questions at the booth especially here at OpenStack We are purely Primary storage for running virtual instances. So we are not an object or file storage solution But because of our express purpose-built design, we have a lot of benefits To offer there now. This is something that enterprise companies Experience a lot and as a cloud user or cloud architect. You may not have these direct discussions with the storage administration team But believe me there are there are issues that that are that are happening there in terms of a simple requirement for 20 terabytes of storage to to give to a cloud environment and OpenStack On the storage side the admins have questions about you know, how much do you really need? What are the configurations that you require? What are the workloads going to look like? You may hear things like raid levels or tier one or tier zero or stripe width or protection policies these are all legacies of Architectures that were designed for physical workloads, but they have real impacts on storage performance Storage predictability storage scalability Even in a cloud environment now Cinder and other other solutions in OpenStack Are geared to try to to hide some of that complexity from the from the cloud user But again, they still have real impacts on the performance of virtual machines and virtual instances Running in any cloud including an OpenStack cloud and that's because fundamentally most architectures for for storage were designed 20 or 30 years ago with a one-to-one relationship in the physical data center between the server that was running the application and The launder volume that was holding the data as we move to virtualized environments We took advantage of shared storage to provide the motion and other abstraction benefits But we ended up with something that some folks call the IO Blender where a container that was designed for To handle one application in the physical world is now running 20 50 100 VMs. They're conflicting with each other you're causing noisy neighbor effects and What's really needed as a move to Clouds as you move to the cloud data center VM aware storage Aligns the applications back to where there should be which is storage. That's fully aware of the virtualized environment So in terms of where Tintry and OpenStack are today Tintry supports OpenStack today with VMware integrated OpenStack or VIO We also support OpenStack with the native NFS driver for Cinder and what we're going to be previewing today and Releasing later this year is a Tintry driver for Cinder that brings our VM aware benefits into an OpenStack environment And why is that important? Again in conventional storage You've got this messy connection between the lones and volumes that are on the arrays that you're using for your primary storage and the hypervisors and Cinder does a lot to simplify the connection from an API standpoint From the hypervisor management or the cloud management system to Cinder But underneath the covers the array is still using lones and volumes. It's still complex. You're having those unpredictable noisy neighbor effects And what Tintry does is really finish that alignment at the per VM level all the way through from OpenStack at the management level through Cinder down to the storage objects as they're being managed And this gives you some really great benefits in terms of out of the box performance Isolation being able to see latency end-to-end from the hypervisor through the network through the storage and even Looking at that over time. We all know that cloud instances are not as disposable as we would like them to be And so we're being asked to troubleshoot problems with virtual machines We can give you full latency views of everything that's on a VM because we're capturing everything and doing everything at the VM level and Finally of extreme importance in private cloud use cases is this concept of a true VM level QoS Where every VM down to its VM DK objects or its Q-COW objects are being Scheduled at a at a per VM level and this allows you to set true tiers of service to guarantee Performance so in a private cloud environment having users be able to choose a tier of storage where they have a guaranteed max and min a gold silver or platinum Tier of service providing some visual updates is in terms of what is the right Policy for a particular VM and as a cloud service provider Being able to see the components of latency including the effects of a particular QoS policy So you know if a VM is suffering performance issues is it due to it being on a Level of storage performance. That's too low. Say the bronze level of service Or is it a different issue in the environment that needs to be troubleshooted and we can give this this benefit to all of our customers in terms of In terms of that visibility We've been doing this for four years with traditional virtualized environments and we're bringing all of this technology to open stack so without further ado and Just to as a reminder, we will be doing a raffle For a drone at the end of the presentation. So if you haven't entered already Monica and her team are Collecting those entries, but I'm going to turn it over to Chris to actually show you Tintry and Cinder integration Alrighty, thanks Chuck So let's actually see some of this in action Got a couple things pulled up right here What we're looking at here is the the Tintry native interface for one of our VM store appliances We actually have it set up as a data store for this rail OSP 6 instance And what I wanted to do is just kick off a new instance And we're going to see how that flows through uses our Cinder driver and then we can actually see that that instance natively at the storage level on our appliance So let's say we're everybody should be getting excited about the drone So let's call this drone We're going to let's just create one of them And one of the nice things we're able to do is we can be a back-end storage source for both glance and Cinder and Nova so when we create this image We're actually making a file system local copy and that means that we can do those copy operations very quickly So that's we're just doing a single instance in this case But you could provision many instances very rapidly and we're offloading those at the file system level Let's select our sent us image and we'll hit launch and Cross our fingers here And so on the back end that's provisioning right now We can see the volume associated with it right over here And I'm going to switch to the the admin view So we can actually see the driver. Oh, sorry See the actual driver that was used to create the the Cinder volume So let's you can see here. We had some previous Volumes that were created with the NFS driver. The one we just created was use the Cinder the Tintry Cinder driver and so if we jump back over here to the the Tintry interface we can see Directly on the storage side. We have a view of each VM that's stored on us or instance that's stored on us as this data store We can do that for Open stack VMware environments hyper V rev KVM And we can do those all at the same time. So if we refresh our view here, we should see our new VM Give that a second The suspense of live demos I know it's always fun What well, that's refreshing here. I just want to show another example We have quite a few more systems loaded up on this environment when we connect VM store appliance to an open stack environment the integration really couldn't be simpler All you do is basically give us a way to connect into your environment. You give us a keystone Endpoint and credentials tell us what mount point within the Tintry file system. You're going to be deploying those Those volumes to and you tell us what region we're a part of and that's all we need And so you can see we're actually simultaneously connected to two totally separate open-sac environments and Simultaneously, we're also connected to a v-center environment So if you're transitioning from a Virtualization environment to a cloud environment we can be kind of a platform to help you make that transition from a storage standpoint So let's see if our our new VM is here So we can see it right there We we saw those files on our file system We made a call back into Nova to find out what that those what VM those files corresponded to and from this point forward You'll be able to manage storage for this instance as a VM not not in terms of lones or blocks Or files we do that management at the VM level So what can we do with that? So on this system? We've got a few more volumes and instances provision we can actually see what are the highest IOPS instances in the environment We could actually pull up Graphs of visualizations of the storage for individual VMs. We can do the same thing we can look into performance for individual volumes We could actually even apply QoS at a VM level so pull up this guy doing about 2200 IOPS and if we actually drop into this QoS mode we can see That this this particular VM doesn't have any QoS thresholds configured If we want to we could drop him down to Say 1500 IOPS up not thousand and So as soon as we do that we actually see these sliders appear so Very easily we can set the the min and max IOPS at a level that makes sense for that VM So you're not sort of blindly setting these levels without knowing what effect that's going to have on the VM And if we look at what effect that has on latency we'll actually see that the Throttle latency that we inject to modify the the storage IOPS if your VM is bursting above that max We actually have to slow him down by adding latency artificially So what you'll see is that that the IOPS are coming down and the throttle latency is going to be coming up So we're bringing that VM into compliance with with the max IOPS levels that we set And we can actually look at that kind of Give you deeper visibility into what's going on with your storage all the way up at the array level So what we're seeing here is a visualization of latency across the whole array and you can notice we've got a huge spike here We can actually drill down into any point in the graph You can say if you're having latency two days ago You can find that point in time and we'll actually tell you what instance or what VM was causing that latency at that time And we'll break it down into latency coming from the host coming from the network coming from a ray level contention The flash or the disc within the array. So really giving you a lot of visibility We're excited to be working with the open stack community and bringing a lot of this The visibility into the storage level into larger higher-scale environments in open stack great And without further ado, I think we've got a drone to give away. Yes, we do Let's just pull that up real quick Any any last entries before we do the drawing? Your chances of winning a drone are very high right now half the people in the audience are our tentry employees And they're not eligible. Yeah Couple more are coming almost there. All right Are there any questions from the audience members about anything that we talked about the the tentry VM store the per VM Performance and management capabilities that we offer or anything. We're doing an open stack. All right We're just here for the drones and that's okay. I Got a prop. Oh good. It was supposed to be red, but I think this one is green All right, do you want to do the honors? Sure? Yes drum roll please All right We have Jason Baker Jason Baker All right, come on up the drone is yours great All right, we we're gonna take a picture right here. If you don't mind if you don't mind He's quicker on the draw You get it great. Thank you. Thank you, sir All right. Thank you. All right. You have a good show everyone