 All right. Let's get started. First, thank you very much for coming. I'm also very happy to be here to attend this summit, since Vancouver is the place where I started three years for my graduate degree. So that is another excuse for me to attend this summit. Today, I'm going to talk about the challenges for open-stack-based hybrid cloud deployment in China and the solutions that we come up to meet the challenges for deploying open-stack-based hybrid cloud to enterprises and cloud service providers in China. First, I'll quickly go over what is the reality of the cloud marketing in China and what is the momentum, the growth trained for open-stack-based cloud, private cloud, hybrid cloud, or public cloud in China, and give you a brief introduction of who we are, what we do, and what is our positioning, and our business model, et cetera. And then now, go over the details of the challenges, the problems that we have to solve when trying to deploy to help enterprises to go to a cloud, open-stack-based cloud. And then I'll go over the details of what we did, what kind of solutions we provided and functionalities we add, and the pinpoints that we resolved for the customers. And lastly, I'll go over a deployment case and tell you the background of the project, the customer, and our solutions. And if you have time, I'll do a quick live demo of our public cloud service site here. All right. So this is a rough estimate of the cloud market share size estimation for China. You can see US, of course, the biggest market in the world, and more than 50% of the market share. And China, surprisingly, the market share is in the lower single digit, believe it or not. As the world's second-largest economy, its market share, the cloud, is actually really its infant state compared with the United States over the rest of the world, for that matter. And I have two slides of some interesting data, researched down by IDC this year, shows you the cloud market, the trend, the cloud management software share estimation, et cetera. And this slide shows you some interesting data and their estimation for the cloud management software in China. So the inner circle shows you the current market share of VMware-based solution, Microsoft Azure Service Pack, commercial OpenStack, and a community OpenStack release. So the dark blue is VMware. And the light blue is Microsoft. And the gray and orange is the OpenStack. As you can see, so for the private cloud sector, VMware is dominant. It's more than three quarters of the market. And Microsoft is gaining a lot of momentum and traction. And it's about 13% while OpenStack here, the commercial version and the community OpenStack is fairly small, so 6%. And IDC, based on their estimate, based on their survey and the estimate that 3% is from now, OpenStack will be almost 20% here. And Microsoft also will increase their share while VMware will be the biggest loser. And it's market share will be from three quarters to about 60%. And also the research looked at the cloud service market area, the IDC operators who wanted to adopt a cloud transformation and started from, transformed from providing the rack and space to cloud service. And the telco service providers. And there are some enterprises who also would like to provide community services, the some vertical clouds, community clouds, those kind of things. So you can see here currently, VMware is about 30% of the market. And three years later, it will shrink to a quarter. While the OpenStack market, you can see first for this market, the gray is self-built. Because the cloud service providers, they have a lot of resources. They have the technology and the expertise so they are really built. And the rest of them, they either use VMware or Microsoft or OpenStack. Again, in this sector, we see a huge increase three years later, based on IDC report, estimation, from 6% to 26%. And there's another way of slice this data, excluding the first year CSPs, like Baidu, Alibaba, and the AWS, et cetera. You can see the OpenStack market share will grow more, excluding the first year CSPs. Now, there is another interesting piece of data is the user survey. The pinning of which cloud management software do you prefer? So when asked this question, 20% of the respondents says it's a commercial OpenStack. And almost 30% said a committed OpenStack. Folks believe, oh, we have in-house IT resources and we can do it ourselves. And 30% said we cloud, VMware still has some big mind share. And about 20% of them select Azure, Microsoft Azure. So the observation or the conclusion is that you can see 50%. 50% of the folks being asked this question choose OpenStack. And OpenStack actually has become the most important cloud technology in the Chinese cloud market. And who are the big players, the super users of the OpenStack technology in China? Similar to what happened worldwide, there are financial institutions, financial service institutions. There are auto manufacturers. Of course, the e-commerce companies Suning and Jingdong and a lot of internet companies like Siner, Citrips 360, NetEase, et cetera, and other education institutions. Based on our feeling of the pulse of the market, of the enterprise market in China, we see a lot of CIOs and CTOs, they already selected their cloud platform technology. That's OpenStack. However, there are three major pain points for the enterprises, for some IDC service providers who want to transform the cloud. The three major pain points are a lack of cost-effective, stable, reliable cloud platform. Now with the event of OpenStack, this pain point is not that big. And the second and the third pain points are the lack of DevOps resources. Cloud is different from a lot of people who say, hey, OpenStack will be the future Linux for cloud market, for cloud service platform. But I don't think that is very accurate, because after all, Linux is a piece of software. While OpenStack is the integration engine, its end product is a service delivered to its users. So you'd have to have people maintaining, fixing bugs, maintaining the clouds, and delivering services to end users on a 24 by 7 by 365 basis. So most enterprises and CSPs, IDC, especially IDC operators, they don't have the DevOps resources. And we are funded to solve these pain points. So the team of Ketum Cloud come from the first public cloud builder and operator in China, the Grand Cloud. Our team has five years of experiences building cloud from scratch. And also we have four years public cloud operation experiences. So last year, we secured VC funding from two renowned VC firms in China, a China broadband capital and GoB partners. And also we received strategic investment from Cisco Systems. And our positioning on the business model is that Ketum Cloud is a hosted hybrid cloud service provider. So this summarizes all what we do in terms of our product and services and in terms of our business model. So our product is OpenStack-based hybrid cloud. Hybrid cloud, I mean interconnected public and private cloud. To support this product, to support public cloud, you have to have a billions. You have to have a reportings, online payment, charges, those kind of bells and whistles that are needed for public cloud, for enterprise or for CSP to provide public cloud services to their end users. And our service model is managed cloud services. So that is our positioning and the business model. So now, what are the challenges? What is hybrid cloud? What are the challenges of OpenStack-based hybrid cloud deployment? Why hybrid cloud? Of course. So the most common definition of hybrid cloud is interconnected public and private cloud. However, there are some other types of hybrid cloud definitions in China, quote unquote. The mixture of OpenStack resource pool, resource cloud with VMware, they call it hybrid cloud. The mix of OpenStack cloud with Power VM or Power VC, they also call it hybrid cloud. In a sense, they are hybrid cloud, different cloud technologies. And also, the bare metal plus OpenStack cloud. And oftentimes, when we work with financial institutions, we found that these customers, these enterprises, they have all of them. They have bare metals. They have a requirement for us to provide them with technology that can manage bare metal pools. They can manage existing VMware investments. And also, they want us to be able to manage their Power VM through Power VC. And they don't want to put everything, all their acts, into the public cloud, especially in China and Asian Pacific countries. Because these guys are worried about their sensitive data. There are critical business applications to be run on a public cloud managed by e-carnals provided like Alibaba. Especially, I met a lot of CIOs and CTOs of the enterprises in China. They say, hey, I will never trust the Ali's Cloud in China. That's the biggest public cloud service in China, not like AWS here. In the US, AWS has had a lot of certifications and gaining a lot of momentum in the federal governments, but not in China. All right. So what hybrid cloud deployment does enterprises really need? So number one, they need a very powerful, for public cloud, they need a very powerful user-friendly tenant self-service portal. So the current horizon provide a good foundation for that, but that is not enough. And oftentimes, some folks, they just use our engines. They build their own portal enterprises and CSPs. The second one is for providing public cloud. You need to have buildings reporting, as I said. And also, you need very competitive ICE or PAS functions that you can compete with AWS, with other public clouds like Ali Cloud or Tencent Clouds. That is, you need a powerful, a cloud admin, a boss system, a business subsystem, or operation subsystem. And the other challenges is what I have just described. They need to manage multiple clouds. And they want us to provide an open stack that can manage the VMware pools, the Power VMware pools, and also bare metals. These are the advanced functions that OpenStack Community Release has not implemented yet. The auto-skilling of VM numbers in an elastic load balancer. The live scale up, increasing the CPU and memory of a virtual machine without rebooting it. We call it live vertical scale up. The ruleback of the VMs OS to any prior snapshot. To ruleback, I mean ruleback, of course, OpenStack provide the snapshot functionality for VM OS for volumes. But developers, what they want is not just a snapshot. They want the capability to rule back their OS, their volume, to any prior snapshots that they have taken for their projects. On and on and on. I'll go over some of this. I have quite a few slides describing, actually, these are the functions that we enhance, we implemented on top of OpenStack. Based on the requirement, when we deploy OpenStack-based hybrid cloud to Chinese enterprises and the cloud service providers, all right. So this shows you the OpenStack as an integration engine for multiple cloud management. So now let's talk about what we did. This is core. Everybody is very familiar. There are three tiers, the UI tier, which is horizon and some command line interface command line tools. And the middle layer is the Alexis services, the novice, Cinder, Swift, and Neutron, et cetera. And the third tier is the share services, the images, the ID management, logging, orchestration, et cetera. As I said, number one, we need a very user-friendly, the cloud platform that gives you the taste of Amazon or other major public clouds. The functionality-wise user experience, the whole nine yards. In addition, we need an admin portal, an engine that allows the cloud admin from a CSP, cloud service provider, or from enterprise, to be able to manage that. That includes, of course, configuring the resources in the OpenStack, which part of the OpenStack admin functionality has already horizon. But more, they need to manage their asset. They need to know the capacity utilization, the monitoring, the billing, the pricing modules, and all the bells and whistles that you need, especially for public cloud services. And they need, as I said, advanced ICE features that allows them to compete with private solutions from WeWare, with public cloud services, functionalities from AWS all year or the rest of them. And lastly, they need billing, service package management, online charge and payment, account order management, et cetera, et cetera, to actually allow them to run their cloud services to their end users. And that is what we call Ketum Cloud. So what we do, we did a lot of addition. And some of the folks, some of our competitors, what they do, they do subtraction. They don't even support the functionalities OpenStack already provided. And we support all the functionalities that OpenStack provided to our end user, and we add more to meet the challenges and the requirements that they have. So this is another diagram shows you what we did. The red ones are the OpenStack modules, the keystones, the new projects, and all the projects that provide the great functionalities. As I said, we need a tenant portal. We need an admin portal, the BSS and the OSS. And we need all the rest of the shared services that allows you to actually manage it, provide the services, be able to give the user the visibility of what is happening in their data center on the server. And you can see there are other factors that we need to take into consideration to deploy Hybrid Cloud, OpenStack Hybrid Cloud, to enterprises in China. I guess I believe here in the States it will be the same thing. Because oftentimes enterprises, they were already investing in clouds. Most of them, probably they already have VMware implementations. And that is especially true for the financial FSIs, financial service institutions. And all of the banks, insurance companies, and security companies we talked with, they have VMware, they have PowerREM, and they have bare metals that they want you to manage. Because oftentimes some of the applications cannot run on virtual machines due to various reasons. While these virtual machines, while these bare metals has to be on the same network or virtual or physical as the other tiers of the application, the app tiers of app tiers. So that presents the challenges for you, for us, to manage the bare metals and also put them on the same virtual subnet as the rest of the virtual machines. So we have to be able to manage the PowerREM pools, the VMware pools, the bare metal pools, and utilize all the existing fiber channel equipments from various vendors. If it's a new deployment, they just buy the commodity hardware, the servers. They don't have any of these troubles. All right. So this is a quick, if I have time, when I have time, I'll show a live demo of this. This is our horizon. You can see the overview of your resources, the compute, the databases, networks. And this is the topology that we enhanced on top of. Basically, we built this tenant portal from scratch using the OpenStack APIs. And this shows you the virtual routers functionalities. Does you have a VPN gateway? And what kind of network is connected, et cetera. Now, there's some details of what we really did. So this shows you the functions that these OpenStack project modules have, you know, Vanuva, Glass, Cinder, Swift. And the comparisons of the releases from the community and the solutions from Red Hat, Morantis, IBM, and HB. And I also compared it with the public, the leading public cloud service providers in China, Qin Cloud and Ali Yun. You can clearly see what we did. First, ruleback of VM, as I said, to any prior snapshot. Auto-skilling the number of VMs in your elastic load balancer, live scale up of VM, that is actually a requirement from a financial service institution. They said, hey, I don't want to stop my application. I want to add CPUs and memories on the fly without shutting down the VM, their application. Resource recycle being, resetting password, et cetera, et cetera. These are all the features that we added on top of OpenStack. For Glass, we added bring your own image. That is, you can create an image from your application in a physical server or another cloud. And you can just upload it to our OpenStack cloud. And that cloud will become your own cloud, your kind of a system cloud. Then you can use it to create virtual machines. And that is really facilitated and speed up your cloud migration from either VMware or Cloud Stack or other public cloud to your own private cloud. In the neutron part, what we did, we added a lot of interconnected abilities. OpenStack support IP stack VPN side to side out of box. It has all the other good stuff. Galaxy IPs for ELB, for virtual router, firewalls for V routers. What we added, we added GRE tunneling, L3 GRE tunneling. We added a host client to site VPN services over PPTP and open VPN. And also we added managing bare metal. The key here is on the same virtual network as the virtual machines. That's what we did. All right. To summarize, we did a lot of work enhance the VM functions, the auto scalings, the live scale ups, et cetera. And this is auto scaling. With heat, you can actually scale the number of VMs in the ELB group like that in a single cloud. So you can come up with a policy to increase the number of VMs based on the CPU, the memories, or any variables, any parameters that you can monitor. And you can decrease this also by setting up some of the triggers, OK? And this shows you the monitoring of one ELB. It shows you the seesaw of the CPUs of this ELB group when you increase or decrease the number of VMs. And this shows you this diagram of the live scale up. You can on the fly, without shutting down the VM, or your applications, and you can increase your CPUs and memories. Ruleback of virtual machines OS snapshot. Bring your own image, BYOI, which is critical. What we did here, actually, we added two APIs, two glass, import and export. So what we did here, actually, is we automated what the Cloud Admin did with their image store. And we implemented the user portal that allows the tenant to do what a Cloud Admin can do. So they can actually create an image, an import. Basically, they can add any kind of images, system images, to our OpenStack Cloud. This shows you a case where the user imported DSL, damn small NX, is only 50 megabyte. And this shows you the virtual machines that are created using their imported DSL image. It's quite different from the flavors of the red hat or Ubuntu of the virtual machines, you can see. Now, the advanced testing functionality that we added, like all three GRE tunneling, site to host, and VPN services over OpenVPN and PPTP, the cloud bursting from one cloud to another. So this is the IP stack VPN that's supported by OpenStack Outbox, and this is what we added, L3 GRE tunneling, allows you to connect one cloud to another if that cloud support GRE VPN gateways. Cloud bursting, a lot of people talk about cloud bursting. So your private cloud has limited resources, have limited elasticity, and during peak time or promotion, you want to burst your applications to the public cloud, where you have a lot of capacity, resources, elasticity, et cetera, et cetera. We actually implemented this with AWS. So we got an AWS account, we created a VPC with a subnet, and we connected these two clouds through IPsec, IPsec VPN tunneling between these two cloud, and we set up our cloud auto scaling, the ELB. Now you have extended ELB over the two clouds, and you can scale out your VMs to the new ELB group in another cloud, in AWS. And when the load comes down, you scale it back. You kill the VMs in AWS first. Scale back your application to your own cloud. That is, we call it cloud bursting, cross-cloud auto scaling. And this figure shows you the VPN from host to site VPN services over PPTP and OpenVPN. To summarize, we developed based on OpenStack. We developed, enhanced the platform so we can be deployed as a public cloud services for the CSPs, cloud service providers, IDC, SPs. And you can also deploy it as a private cloud. And we can connect it with the site-to-site secure VPN services. And we also can connect to their customer office. Our third enhancement is we prioritize our boss system, our engine. So you have account management capabilities, IDC asset management, deployment, monitoring, online invoice, integrated support ticket, the bells and vessels that a service provider, a public service provider need to provide public services. And this shows you the boss system, the key engine, the pricing. OK, I think I'm kind of short on time, so I'm going to speed up a little bit. The third improvement what we did is we must be able to support the complex deployment environment in enterprises. As I said, they want you to connect their private cloud with public cloud. And they want you to manage bare metal, manage VM and pool, managing power EM. And we can do that, because when we are recenter, open it up their API, recenter 5.5, open up all the API, restore API from OpenStack. Actually, you can add a resource pool. So when the user say, I want to go to the VM or pool, you can direct the user to that VM or pool and create VMs in that VM or pool. You don't have to do with anything. You don't have to remove the recenter, the resource management software for VM or pool. You just leave it there, and you can manage it. And that's the requirement. The same thing for Power VM. And we did that. I'll give you a couple of snapshots that shows you so here. So when the user select the image, created their own provision, their virtual machine, you give them a choice of the local resources using KVM or the VM Ware. So they show you the images that are available on the VM Ware pool. We can do the same thing for Power VM. Most importantly, for bare metal, Power VM and the VMs in the VM Ware resources, we are able to put them on the same network, to allow them to join on the same subnet, the virtual network. I think I'll give a quick demo of five minutes. So all right. This shows our key cloud. You can see you have the compute storage. It's a little different from what you see in the horizon. It shows you the network. So this is the virtual machine, the topology. It shows you the virtual router, which is where it is on. This one has a virtual VPN gateways. It has a floating IP, a public IP. This is a SUSE instance. Shows you the image, the network it is on, et cetera. So overview shows all the resources, the recent activities. Compute shows the instance images, recycle bins. Recycle bin is the place where we store the resources you delete for two hours. And after that, it's permanently deleted. NOVA has this API called soft delete. And it also has a force delete. Basically, if you delete by mistake, it will be here. The other place that you can see all your resources is here. It's a network topology. It shows you what virtual routers you have and what you have on these water routers. Do they have a gateway? Do they have a floating IP? And this shows you your subnet, your virtual subnet, all the overly virtual subnet for this tenant. And also the virtual machines on these subnets. All right. I'll do a quick showing of the, this is ELB. And this ELB has two members. And I set up auto scaling on this ELB. It doesn't need to add one VM when the CPU is greater than 90 and delete one when the CPU is below 75. And what you see here is you have this seesaw CPU pattern. And that shows you the system auto skills, auto add, and shrink the resource pools based on the load. A quick look at our engine, the cloud enemy engine, an enterprise, or a CSP. They need this. They need to know what physical servers they are on, the CPU utilization on this server, and how many VMs on this server. They need to know the asset, the servers, the switches, the IDCs, what resources you have. Their IP addresses, all the information you need. And they need billings. So resource settings are what admin of the horizon already have. We added asset management, billing, support, and statistics. So this is a pricing engine. It shows, first, you can design your product. What product you want to charge. The images, the bandwidth, the volumes. And then you can decide what kind of price you want to charge. You can change it here. And also, you have special prices for your promotion. And this will all write the prices you set in the normal engine. And last, but not least, it gives you the capability to see what is going on, the visibility of the resource utilization in your data center. So this shows you the number of paid VMs, the CPUs on these VMs, your daily charges, the new users, the daily active users, and the flavors of your virtual machines. And user statistics, since we were the first public cloud service provider, so we know a CSP for a public cloud. You will need this information. Your paid subscribers, your registry users, consumption users, your daily active users, et cetera, et cetera. That's it. Thank you very much.