 Hey, thanks, everyone, for coming for my session. It's the last session for the day. I hope I can keep you engaged with my presentation. So today we'll discuss tapping into the strengths of hybrid cloud. Over the last few years, I think this hybrid cloud is something that is gaining momentum, right? So we had a public cloud and private cloud, and now it's a hybrid cloud. Hybrid cloud is a combination of two or more clouds. And let's take a look at what the, oh, there you go. OK, so in this session, we will take a look at what the introduction to hybrid clouds. And we'll also take a look at what the enterprise IT composition. And then we'll see what are the key enablers for the hybrid cloud, right? And then we at Triller Data how we battle multiple clouds during our testing and development and deployment phase. So let's start with the introduction to hybrid cloud. So how many of you are thinking of deploying hybrid clouds in your enterprise? OK, so do you have any future plans in the two years down the line? Does your IT, looking, leveraging public cloud some form or shape? OK, so based on some stats that we picked right now, the hybrid cloud, Gartner says, ITs will leverage the hybrid cloud in some form or shape. Almost like 50% is going to be like hybrid cloud, right? And other research, it varies from 40% to all the way to 80%. The composition varies. But the key point here is, in some form or shape, the hybrid clouds are gaining momentum and they are here to stay. Now, when I look at the hybrid cloud, I'm only talking from the IT point of view. If you look at Google Mail, Gmail, that is some kind of cloud that you are using. If your enterprise is using Gmail plus something else, it could be some form of hybrid cloud. But we are only looking from enterprise workloads point of view, how we can leverage the on-prem and the public cloud to basically effectively implement a better IT. Now, what is a hybrid cloud? So from a simple definition point of view, the hybrid cloud is a combination or loosely coupled two or more clouds that acts as one cloud, right? And typically, when people say hybrid cloud, it is like you have something on-prem. It could be VMware cloud or OpenStack cloud. And then you combine that with one or two public clouds. Essentially, the idea is to leverage the capabilities that various clouds offers to implement a better IT for yourself. So obviously, if hybrid cloud is done right, it definitely enhances the reliability of our IT, the flexibility of the IT. But more importantly, it can decrease the cost. Like cost of ownership can be enhanced by adapting a hybrid cloud strategy. Now, so the other research says total cost of ownership will be reduced by 54%. That's a big number. Obviously, when you look at the hybrid cloud and you look at the public cloud composition, the thing that we are looking is a utility cloud model. And most of the time, we don't need dedicated resources for the tasks that we are running in our IT. If somehow we can leverage the utility model that the public cloud offers and then only pay as we go, so it definitely reduces the cost of ownership. And also, it enhances the operational efficiency because you spin up the resources and spin down the resources when you don't need them. And the rest of them will fall apart. It facilitates the innovation, meets customer expectations more readily. Obviously, there are very good benefits of employing or adapting hybrid cloud strategy. But what does hybrid cloud strategy really means? If you talk to 10 different people, you may get 10 different answers. And depending on what the vendor that you are talking to, he may spin his own version of what the hybrid cloud is. But if I look at what kind of deployments of hybrid cloud are there, I can boil down them into two categories. One is you have the unified management. So that means you take one or two clouds. It could be on-prem and then the public cloud. And then you layer a unified management layer on top of this cloud so that it can provide you a single pane of glass for your management. And it helps all the details of your individual clouds within the hybrid cloud. And then if you are spinning up some workload based on what is needed from the workload and then analyze what the capabilities that each of these clouds offers, it can determine what the best cloud that best suits your workload and deploy the workload there. And also, they help you provide some governance around who is deploying what in your organization. So this is one of the very popular way of deploying hybrid cloud. Scalar is the product that seems to be very popular. Open source project, but there are other projects out there that can help you implement this unified management for your hybrid cloud. So that's all good things. But what are the disadvantages with this one? So it's still like you have a unified management and you can basically deploy these workloads either cloud one or cloud two. But then you need to plan a complete life cycle of your applications. It's not like you basically identify the cloud and then deploy the application. But what about the rest of it? So here you are essentially building silos. You may have three or four clouds within your hybrid cloud. So you choose one cloud for one workload, one cloud for other workload, and then what? You are basically stuck there. If you choose a cloud one and then you deploy one application, your application purses there. And there is not much synergy that's happening between the clouds. So once you deploy your application and then you want to leverage the capabilities that are available on a different cloud, but it may not have access to the application data so that you can leverage the synergies between the clouds. So this is one popular deployment model for the hybrid cloud. So the next one is the replicate data sets. So you have your on-prem applications and then you have a storage. And then the storage can replicate the data sets, whether it's a LANs or files, back into the public cloud. So you have application data sets available in the public cloud in case you want them. This provides some kind of security and then maybe availability of your data sets. But this one also comes with certain disadvantages. For example, if you are replicating block storage. So essentially, each block is replicated to the public cloud. Now, how do you lay out your application knowledge on this set of LANs? So we don't know if those LANs are used by two VMs or 10 VMs. And then we don't even know whether those LANs are formatted with certain file system or certain applications on the LAN. So unless you have how those LANs are used on your on-prem, you can't make sense of what is being copied onto the public cloud. So now if you look at public cloud and private cloud, people did a lot of research on what makes sense and what doesn't make sense. But overall, I think your hybrid cloud implementation should have a way to utilize the utility model that public cloud offers. Abundant supply of compute and storage resources. You spin up the resources as you want and then spin down the resources when you are done with this. So your hybrid cloud strategy is not utilizing the utility model that the public cloud offer. Your hybrid cloud, you are limited by what you can achieve out of the hybrid cloud. Now, so the two hybrid models that I talked about. One is unified management layer. And the other one is some kind of gateway so where you can replicate your data sets into the public cloud. Those are well and good. But they almost looks like a greenfield opportunities. So if you are starting from clean slate and you are building an IT from ground up, well, these two models make perfect sense. But what about the existing IT? It's not like IT, it's been there for years. And then you have a lot of applications that are already running on enterprise IT. How do you basically deploy hybrid cloud strategy for your enterprises? So this is something that we'll explore in the next few slides. Now, when you look at the enterprises, we do a lot of things in IT on a daily basis. One is the functional requirements. Essentially, these are the business critical processes that need to be up and running all the time. If anyone goes down, that's not good for the business. So it could be your HR payroll or CRM system, email server. So all these are business intelligence or analytical systems that are up and running all the time. And then the next one is non-functional requirement that typically the IT users don't see them, but you still need to execute those things all the time. For example, you need to make sure that your applications are performing at the peak scale so that all the SLAs are met. And you need to implement some kind of BCDR. And then you need to ensure that the data security is preserved and the network security is preserved. And when you are deploying or even you are doing the tech refresh, you need to make sure that you are testing your application. So there are a lot of things that goes behind the scenes. Those are non-functional requirements. Now, let's look at what resources are needed for the functional requirements or some non-functional requirements and then see how we can map it to the hybrid cloud implementation. So when you look at the non-functional requirements, obviously these are dedicated resources. You got to have those applications up and running 24-7 and you probably need the best of the breed hardware resources available for running those things. These are business critical applications. So these are pretty much belongs to the on-prem if you are basically going with the hybrid cloud approach. But then when you look at non-functional requirements, these are not a critical path of the business, but you still need to have those things to have the IT. Those are important for the IT. And some of them are like network scanners. Maybe you need to run some virus scanners or the network security applications to make sure that you have the right security in place. It may be test out. And for these things, you don't need dedicated resources. And for these kind of requirements, you can definitely leverage the public cloud so that you can basically spin up the resource as you need it and then spin down. So you can use that utility model to make your IT more agile, but also most cost effective. So how do we basically dev up these things and then move some of the functionality that you perform on-prem in the IT to the public cloud? So one of the things that we needed is, obviously, access to the same data set between these two requirements, both functional and non-functional requirement. So for example, if you are moving the BCDRR test to the public cloud, so you need to have the right data set from the production application so that your test is effective. So what that means is they need to have access to the same data set that your production is running so that your test and dev can be more effective. So you need to have this layer where this data access flows between on-prem and all the clouds within the hybrid cloud. So I call out like, these are the four. How do you achieve that one? So I called out four key enablers for the hybrid cloud. So one of them is the platform agnostic. So we deploy lots of applications in our IT. But we don't capture these data sets in a way that is consumed across all the clouds. So if you are deploying a hybrid cloud, if you can't consume your data sets that you are deploying on one cloud, on a different cloud, that limits what you can do with the hybrid cloud. The other one is ability to mobile your data sets between the clouds. So on cloud one, you have the application. But you should be able to analyze and run some applications on the same data set on a different cloud. So you should be able to securely move these data sets between the clouds. The third one, orchestrate. So let's say you have an application that is running on-prem. And you capture the data sets in a platform independent way. Now you can access that data between these two or three clouds. But then at some point, you need to go and realize the data set on a different cloud. You need to re-orchestrate that application on a different set so that you basically get the most value out of your data set. And the fourth one is the single pane of glass. So whether you are building your hybrid cloud with one cloud or multiple clouds, you should not expose a lot of complexity to the end user. If you have the single pane of administration between the clouds is under. So these are the key enablers I would say is important for the effective implementation of the hybrid cloud. So let's dig down deep into each one of these attributes. One is the platform agnostic. So it's very, very important that any data we capture in one cloud should be able to consume in a different cloud. But getting a platform agnostic or platform independent way of capturing the data is not that trivial because the current IT uses lots of different platforms. So each capture looks a lot more different. And also when you look at how we basically build the IT over the period of time, there is not much separation between the application and then the platform. So we load our application server with a lot of filters and agents to realize the non-functional requirements. So it basically bloats our IT and then also makes our IT a lot more complex. So that is one of the reasons why it's very, very difficult to yank our applications in a platform independent way so that those applications can be mobile between the cloud. So one of the basic things that we need to do to leverage the hybrid cloud is stand days on a platform like obviously OpenStack. So that at least the on-prem can run a standard platform and all the applications that are running on this can be platform independent and provide the mobility between the clouds. And also, you need to have access to the same data between the clouds. So that means you need to capture your applications in its entirety and also capture them more often. So that all this data can be analyzed or used standard tools, maybe Linux tools or some standard off-the-shelf tools to implement various functionalities in various clouds by leveraging the underlying capabilities of the cloud. Now, the next one is the workload mobility. So according to Cisco study, 64% of the respondents say they need to have an automated way of moving the workloads between the clouds. So the second factor, the workload mobility is also important. So we talked about platform independence and then now we are talking about the workload mobility. So how do we achieve the workload mobility? How do you basically move the data between the clouds? So traditional storage like NFS cannot scale between the clouds. So the best way to move the workloads between the clouds is the cloud storage. For example, S3 is the most popular. And then other cloud storage can also be used to basically persist the application data but also able to access between the clouds within the hybrid cloud. And the third one, the third enabler is the orchestrate. So if you have some application that is built in one cloud and you already captured that in a platform independent way and you have a means to migrate the data to a different cloud, the third step is are you able to re-orchestrate that same application on a public cloud, on a different cloud? So I'm taking this typical application here. So it's like three VM deployment. It has various storage attached to that one. And then we have two or three networks connected to the VMs. And assuming that this is running on an open stack, and how do we re-orchestrate the same application on Amazon? Fortunately, between Amazon and open stack, the mapping of the resource drives that are available between the two clouds, they match very well. So it's pretty straightforward to take whatever is captured on an open stack. For example, if all the VM images and all the sender images are captured in a standard format like QCOW2, they can be converted into AMI images and rest of them can be mapped from open stack resource types to the AWS resource type and then recreate that application relatively easily. And the third one is, obviously, you need to have one administrative domain for managing all the clouds within your hybrid cloud. So if you have one tenant in your hybrid cloud and you log into your hybrid cloud, you should not feel that you need to log into all four different clouds. And the single pane of glass should hide all the details and provide you one simple interface for you to consume all the hybrid clouds. So we talked about four key enablers, right? It's a platform independent of capturing the data and then able to migrate the data between the clouds. And the third one is able to re-orchestrate any of this application set in any of the clouds of the hybrid cloud. And then the fourth one we talked about is the single pane of glass. So assuming that you got what I was trying to tell you correctly, so one trivia for you, what is the most important cloud in your hybrid cloud? Can anyone take that? So if you can get the hybrid cloud strategy right, my answer is none. Because at the end of the day, your business application is the most critical. If you capture the data in a way that is a platform independent, where you are running your application is completely irrelevant. You should be able to leverage any cloud that is there part of your hybrid cloud. So that is the journey that we need to achieve so that our hybrid cloud strategy can be more effective. Now let's take a look at what we do on a daily basis. Trilure data, we battle multiple clouds, even though multiple OpenStack clouds. To just give a quick blip of what we do, we offer data protection as a service for the OpenStack clouds. So just like any other service within the OpenStack, ours is an add-on service to your cloud. And we provide tenant level backups. And so what we do is we don't just capture your sender volume or the files. We capture your entire application. So if you are running a multi-sender volume application, we capture the entire application environment as one snapshot. And we persist across multiple stories like NFS, or Swift, or Safer, in the future, Amazon Web Services. So the reason why I'm saying is that makes our support matrix a lot more complex. We need to test with different sets of the OpenStack distributions and different releases of OpenStack distributions and then different target repositories that we need to do. So most of the time, these are the common questions that the customer get to ask. If you are deploying OpenStack, you don't really limit yourself to one OpenStack. You may be deploying OpenStack on Liberty that's running on production. Now you are planning to upgrade that to the Newton. So it's not like one fine day you go and upgrade your Liberty Newton. That's not it's going to work. Most of the time, you stand up a different cloud that is running the Newton. And then it goes through some testing. And then you want to take the workload that is running in the production and then recreate that on Newton and test the production and make sure that they are working up to the mark before you fail over to the new release. So this is the common theme we see with our customers. So these are the questions that comes because of that. So they have multiple clouds deployed. How do they migrate the workload from one cloud to a different cloud? Or they set up two clouds and they want to use it one cloud, a DR side for the other cloud. So how do they make sure that the workloads can be re-archestrated? So because the way we capture the workloads in an entirety, we can orchestrate these workloads between different versions of the OpenStack cloud. So our support matrix becomes very, very complex. And not only that, for every time we release, we need to test the migration scenarios between any two versions of the cloud. So it's quite common that we take a backup that was done in Kilo and then we should be able to restore to the Newton. So even though these are all same clouds, even though they basically have the same API set, but it is still a challenge taking the complete workload in one cloud and then restore it and make sure the workload is working all right. Because your resource type may be changed, right? So the Kilo version of the OpenStack that you are running may be running on LVM, and then the Newton may be running on CIF, or maybe the network configuration has changed, or maybe the regions has changed. So there are a lot of things that can change between two clouds. So whatever backups that are done on the Kilo, they should be able to run on Newton. So we do constantly test the migration scenario between various clouds. This can be only achieved basically, we adhere to that four principles, try to be a platform agnostic. So all the backups that we do need to be platform agnostic. So we should be able to take that and recreate that on any other OpenStack. And we should have a common way to migrate these data sets between the clouds, either through NFS, or through SWIFT, or the CIF, and then re-orchestrate. So you need to recreate the entire workload that was backed up onto your destination cloud. So even though it's not the true private to public hybrid cloud scenario, but it is still nonetheless, it's a multi-cloud management. It basically takes the same challenges as implementing a complete hybrid cloud. Okay, so in summary, we quickly went through like what the hybrid cloud is and what the popular deployment models of the hybrid cloud are. You know, the challenges in the enterprise, I think to make it enterprises more agile. So we need to capture everything in a platform agnostic web. We need to be able to migrate the data sets between the public cloud and private cloud. We should be able to re-orchestrate this application. So these are the things we talked about it, to make it IT is more agile and leverage the hybrid cloud, all the good things about the hybrid cloud. So that's pretty much it. Like any questions made sense? Okay, so we'll be doing a quick demo tomorrow in the theater, essentially demonstrating how we capture the workloads in one cloud and migrate to the different cloud. Please welcome, please, we invite you to join there. Thank you.