 Okay. Good afternoon everybody. My name is David Ke and I'm from Shenzhen of China. I'm honored to be here. I will share something about building our financial cloud. I will focus on first three questions. For the fourth part, I will have two experts to introduce and explain the solutions provided by Mirantis. The first question is what kind of clouds are we building? Before I answer this question, I want to tell you the background about our company. Shenzhen Stock Exchange was established in 1990 and it's one of the two stock exchanges in mainland China. Shenzhen SE has 1,700 listed companies with a total market capitalization of 514 billion U.S. standard by June 2015. It consists of the main board, MSC board, Chilex market. SCSE also has a multi-tiered capital market system for issuers at different stages of growth. Its products include equities, mutual funds and bonds. SCSE, Shenzhen Securities Communication Corporation was established in 1993. Its sub-company of Shenzhen stock is changed. It provides communication services including market data, trading and caring for Shenzhen stock market. Meanwhile, it provides a series of IT services for financial institutions. Let's look at the picture on the left. From 26 to 2014, the size of Chinese asset management has increased from 3 trillion to 36 trillion. That's a 10-foot growth in 10 years. On the right picture, the monthly turnover of Shenzhen market has soared from 2.2 trillion in 2014 to 14.7 trillion in May of 2015. The highest one is about 6 times the lowest. As a result of the rapid growth, securities market in China required advanced IT systems with high capacity and resilience. In the past two years, the size of many assets has a constant expansion, while more and more financial licenses were issued. The number of financial institutions also increased a lot. By September of 2015, according to the registration information from the AMAC, Asset Management Association of China, the number of private equity funds has reached 20,000, increasing by 200% comparing to the early of 2015. They have managed the assets of 4.5 trillion. This small or medium-sized organization helps to focus on the business which can show their professional abilities and also the long-term IT to other professional institutions. We see the huge demand of IT systems on the above slide. However, we still have to face two major IT problems in current security market of China. First, the traditional IT framework has some problems such as tight coupling and vendor locking, high risks, low efficiency and slow response to the business. So the traditional IT framework needs urgent reform. Second, lacking preling and co-ordination scattered resources the industry's infrastructure lacks behind. We believe that cloud computing is the most effective technological means to meet the market demand and solve the business problems, especially for its features such as elastic expansion and quick response. However, the private cloud has some problems like small size, technology weakness and slow application migration. It is also difficult for the public cloud to fulfill the requirements of data security and regulation. For the time being, by combining both industry and private clouds, the hybrid cloud structure is relatively suitable for the financial industry solution in China. As one of the main IT provider in the financial industry, the SCCC is now constructing and operating financial cloud. This will be our main strategic transformation. The financial cloud that we are building is an open, cooperative and win-win ecosystem. Based on an open cloud platform, many as with independent software vendors provide a rich application software services, each financial institution can choose from two or more suppliers. This diagram shows the layer service architecture of our financial cloud. Our company mainly focuses on offering the basic services, like ICE, and we work together with our partners. Some ice weights to offer pass and sus, customers can use various service applications they need. This financial cloud has four features. First, open. We use the mainstream open source cloud platform, and our platform is open to all our partners. Second, neutral. When standard stock is changed as our shareholder, the SCCC has enough credibility and the independent third party on business. So there is no commercial conflict between our company and customers. Third, we have professional infrastructure, technology platform and application service. We present our expertise. Fourth, with service experience of more than 20 years, we can ensure the data security for our customers and regulators. Since 2014, we have launched our financial cloud project. We have established several data centers in China such as Shenzhen, Shanghai, Beijing, Dongguan, Hong Kong, Wuhan and some other cities. Our financial cloud has the ability to offer service to customers all around China. I will talk about our new data center under construction in Dongguan. This center covers an area of 130,000 square meters with a 3 billion RMB total investment. The first phase investment is 1 billion RMB with a total capacity of 20,000 racks. The first phase is 5,000. This is a core production center of our financial cloud. Then let's talk about the cloud management system. This system manages the infrastructure of our cloud including computing, storage and networks. It handles all the resource requests from application systems. This system is an essential software component of our financial cloud. Later, I will focus on discuss our choice. After the exploration and practice in the last two years, we have successfully run a series of industry applications on our financial cloud. The first one is a market data cloud. We have established ICE platform in four cities in China. And Securities Company set up their service on this platform. With this service, investors around China can get real-time market data information through internet. With this cloud, the securities company can reduce costs by 30% and this response time of system expansion shrinks 90%. Up to now, the maximum number of online users at this same time is more than 1.2 million and more than 12 million visitors daily. The second one is fund cloud. Our company offers basic ICE and ICE offers applications that mature funds need. Fund companies run the standardized SaaS service on demand. As a result, it can reduce operation costs and save the provision time. The third case is our business innovation for using face recognition. We hope the Securities Company to build an online open account system on the financial cloud. It saves time and reduce costs by using the face recognition technology. Up to now, thousands of financial software vendors have become cooperative partners with our financial cloud. And the ICEV community began to take shape. Here are the logos of some software vendors. They cover the areas of asset management, market data information service, and quantitative trading, and so on. They can provide financial customers with various application software services. We have a face development plan for the financial cloud. The first phase is exploration period. The cloud size reaches 600 nodes or 12,000 vm. The second phase is ecological cultivation reaching to 1200 nodes or 24,000 vm. The third phase is maturity reaching to 3,000 nodes or 60,000 vm. As I mentioned earlier, cloud management system is the core of our financial cloud. I will talk about why we choose OpenStat. We want our financial cloud ecological to be open, collaborative, and win-win. The technology roadmap is consistent with the same open spirit. We want to avoid single vendor locking. When we choose open source software, we could invest on R&D and innovation to optimize and design. Those that our cloud platform can be the requirement of financial customers. We could master the cloud platform technology and enhance our capability to compete in market. OpenStat currently has a largest ecosystem including hardware vendors and software vendors. Compared to other open source cloud platform software, OpenStat community is the most active. OpenStat enables new technologies such as Docker and SET to be integrated fast. Soon our new Dongguan data center will be ready for operation. Then we will have a large number of racks. Large scale, scalability, and high performance is the design goal of OpenStat. This is in line with our requirements. OpenStat has a good cloud platform API interface. Accounting for this, we can build a software-defined data center and create a personalized business support system. OpenStat has released every half a year. The function is more and more rich and perfect. The enterprise can choose the right components according to their own needs. We choose Mirantis to help build our cloud and why. There are a lot of big companies and startups focused on OpenStat private cloud. I would like to share some points. Open source and technology are two of the most important factors when choosing a service vendor for us. Mirantis is well-known in OpenStat community and enterprise cloud market. The design solutions are fully based on customer requirements. As a financial cloud provider, we have a lot of special feature demand. It can always satisfy through customer design. Another aspect that makes us choose Mirantis is that their solution is totally zero-luck. The product can be integrated with various open source and commercial hypervisors, storages and SDN solutions. This gives us the flexibility to build cloud for complex use cases. Well, as the time is limited, I have to stop here. Let's welcome the CTO of Mirantis China, Mr. Zotin, to share his overall solution. Hello. Thank you, David. My name is Tien Zao and I'm the CTO for Mirantis China. This is my second time giving the presentation in the OpenStack Summit, but it is my first time to represent Mirantis to talk about some successful story for our customer. Well, I have been living in the Silicon Valley for 16 years before I took Mirantis offer to relocate back to Beijing, China to help set up the Mirantis Chinese entity and also ramp up a local Chinese technical team to better serve our local Chinese fast-growing OpenStack market. Ever since I came back to China, I have been talking to so many large enterprise customers and been listening to their requirement for building private or like this case, industry cloud platform using OpenStack. I think we face a challenge for most of the foreign company because the OpenStack in China basically the business is going very fast, but all the local Chinese customers facing a big top concern. They are concerned about the local support. So I have been talking to the customer, Chinese customer who not only require a stable enterprise OpenStack software district, they also demand a lot of localized special features and also industry related features just like this case. As we just heard about from Shenzhen Stock Change, actually Shenzhen Stock Change is planning to build a pretty large scalable financial industry cloud. And we have been working with them to make sure the customer success is always our first priority. And we usually help our customer to achieve their business goals through three stages in our project. Merantis usually starts engaging with the customer in a very early stage to make sure that the cloud, their building actually can meet their business needs instead of just blindly following the latest technology trend. And once the project gets kicked off, we usually sit down with the customer face to face and it's senior technical team three to five days at workshop, intensively discuss with the customer about to help them analyze their business needs and also try to understand the design of the customer's existing or future workload. Clear any doubts with the customer about the requirement in security, networking, storage or reliability requirements. Once we kick off the project, we work with the customer to create an architecture blueprint. We call it architecture blueprint with detailed cloud architecture design, including recommended underline network topology, storage design, SDN controller design, or even hardware recommendation list after. And the second phase we call customization and deployment based on the previous architecture blueprint. In case customer needs any customized integration, we help customer to fully automate this work through our few plug-in because we believe the automation is must have for building a large enterprise ready cloud. And once the deployment is done, we also perform a full blown health check automated test script to make sure the cloud is functioning correctly. The final stage, in the final stage for the most mission critical workload, it's very important the cloud is always available. So Mirantis provide the customer different level of technical support, 8x5, 24x7 or even proactive support. Proactive support includes a customer successful manager to make sure customer's goal can always be achieved at any time. So at Sentinel Stock Change, Mirantis was asked to deliver a secure stable cloud platform. This picture basically tells about very high level architecture design in our phase 1 projects. We are using KVN as our hypervisor for the Mirantis OpenStack because we believe KVN is best tested and supported in OpenStack community. Both distributed storage solution staff and traditional send storage are supported to provide a multiple choice for sender backend. Juniper OpenControl is chosen because we believe it's an open source SDN solution ready for large scale secure deployment with some extra features such as service chain. And it is also very well integrated with the Mirantis OpenStack field to provide automatic deployment for SDN solution. We implement high availability in almost in all components in this cloud including but not limited to OpenStack API, MySQL, RabbitMQ, SAF, SDN components or send storage array. We use our two deployment tool field to deploy and manage all components in the cloud itself. The end-to-end process is fully automated. The LMA tool chains in Mirantis OpenStack 7.0 provide the next generation login monitoring and alerting solution for the cloud. And we are going to use Marano to deploy and provide the financial workloads if needed and also using Sahara to quickly spin up Big Data Hadoop cluster if needed. So now I'm going to ask our solution architect Yaguantang to spend a little bit more time to do a quick deep dive in each area. Thank you. Hello everyone, I'm Tang Yaguang. I've joined Mirantis China for almost a year and before that I'm working for Kinanico and also for the OpenStack area. Today I would like to share some details of our solution for the SACC. The first is the Junifor control as the selected SDN solution. Together with SACC we have chosen Junifor control as SDN controller to provide a stable, scalable and secure virtual networking solution. And the Junifor SDN controller is deployed in HRE mode, combining the different controller components on a single node. Also the data plan and the north source traffic in HRE mode are controlled by lowering the scale of your north and south gateways. And the horizontally scalable is one of the major benefits of Junifor control is the independence of the central L3 agent. And running the work we wrote on each compute node. Each compute house is responsible for its own instances, distributed load and solutions to failure domain. So this is much different from the community open source neutron solution. We know that one of the major issues in the neutron so far is that the central L3 agent and we already have a solution called DVR. We have distributed virtual routing, but it requires almost every node to be a gateway. It's very flat. Junifor gateway is scalable through the hardware. And Junifor SDN also provides service training for advanced security services such as the WIT DPI and the virtual firewall. And another benefit is the Junifor SDN solution is also no hardware login. Contrary works with any underlist switches and multiple hardware gateways. This is also one of the requirements of SSCC. They don't want any hardware locked in the solution, including the storage and the networking and also the hypervisor. And Junifor also provides the advanced networking analysis of the whole networking, including traffic inspection and topology visualization. So this is why we are useful for the cloud administrators. And the next is that we not only build the HA solution for the software networking layer and we also build the cloud solution for the hardware. So this is a total solution, not only OpenStack. We have helped SSCC to design the hardware and the rack. So the solution is like that. And each rack we have the TOR switches and the two switches are bonding through LACP protocol and all the TOR to span all the level three. So that we can see from the picture that all the networking traffic is reductant. As a result, so we not only build the HA in the software side and also the hardware side to make sure that the mission critical applications are always useful. And the next is our storage design for the requirements. As we know that for the financial industry, they have some mission critical applications with guaranteed high performance and reliability requirements. For these workloads, we want dedicated sense storage with available multi-links for the optical reliability and performance. The downside of sense storage is cost inefficiency and scalability. Sense storage is posed as a standard backend. But we know that luckily most of the workloads only need high reliability and medium to high performance. So for those workloads, we utilize the SAP distributed storage software solution that provides both reliability and scalability cost efficiently. We utilize two different SAP posts. One is the SAP SAS pull, which is posed as a standard backend and also as a functional storage for NOAA. And in this way we can provide low migration functionality and provide quick backup and recovery of compute nodes. And for the SAP SSD pull, which is only posed as a standard backend, users can create volumes for high performance workloads from this SAP SSD pull. Depending on the workload, users are afraid to choose which SAP backend to use. For example, we could use SAP SAS pull for the operating system disk and SAP SSD pull for the master code application. And the other is the big data on the big cloud. And the established data processing on cloud is 1.0 as a success use case. The one-on-demand automated adobe cluster for the small jobs. We are using OpenStats CoreProject Sahara combining with Cinder's feature. Cinder has a block device driver which can use the physical house disk provided to the virtual house, the VM. So a group of servers are assigned to be a house aggregate and it's posed physical disk to the HDFS VMs. So in this way we make it possible to run adobe cluster on OpenStats with no performance issue. And the next for Morano. Morano is also one of the core project of OpenStats is for the application catalog. For we use Morano to enable cloud administrators to publish various cloud-ready financial applications in a portable cataloged, catalyzed catalog. It can also be integrated with the enterprise PSS system. With Morano we can make cloud in the users to self-service application provisioning, repeatable deployment for port and staging environments, and the standard building blocks for different application stacks. And the last is the LMA. Morano provides a solution for logging, mortgating, and alerting of the OpenStats cloud. As we know that we build cloud not only we finished, we have to do a lot of work to make the cloud work successful. To make the cloud applications works and we have to design a mortgating solution. The LMA is one of the multi solution. This two chain is service-originated, mortgating solution. It's comprised a collection of open source tools to help you monitor and diagnose service availability problems in your OpenStats environments at the infrastructure level. By combining the established search with Kibana, the LMA two chain provides an effective way to search and correlate all service-affaging events that occur in the system for the root cause. And alerting, including logs and notifications from the OpenStats ceramic messaging bus. Likewise, by combining the inflex DB with Grafana, the LMA two chain also brings insightful metrics analysis to visualize how OpenStats behave over time. With these two, we can do log analysis and troubleshooting, and we can also monitor all OpenStats services health status. We can also use this to do more further work, because it formulas all the data and in your standard format you can use the data for other use cases. So this solution provides a way that we can monitor the cloud in a wash-less way. So this is some details of our design solution for SSCC. And for any questions? Firstly, are you posting a copy of this presentation anywhere? And secondly, if you could do this again, what are the one or two lessons that you've learned from developing this FSI cloud on OpenStack? Sure. Let me take this question. Definitely, we will post this presentation also. I think I don't know whether this time we have video shooting and it will be posted online. And definitely, this presentation will be posted on the OpenStack.org, I guess. If not, you can talk to me offline. I can send you a presentation. And I want to emphasize basically this is in the early stage for our phase one project. Right now we finished our architecture design and we actually in the middle of deployment. So actually we are planning to do another sharing for the next year to talk about the lessons we learned and the scalability. Even I think until next April we will have this financial cloud online already. We target to bring it online early next year. So at that time definitely we can share a lot more about the lessons we learned in the deployment stage and also the lessons we learned after we bring it online. So stay tuned. We will share later. Actually I have a question about your LIMA. So you just mentioned you can capture the errors from the logs. I'm curious that you will get error information from the OpenStack log file. For example, the NOVA log file. And show it in the GUI. Yes. So we are collecting all the log files from the computer news and the controller news. And send them to the established search for the virtualized showing. You can select filtering the error messages. It's a way to help the cloud administrator to troubleshooting and to check the cloud health status. Another question is that do you have any additional tools to do that or are you just building your package? We are all using the OpenSource tools. So we are using a collect in the computer news to configure a plugin to collect the files and sending them to an LIMA collector centrally. And then put them to the established search. Okay. Thank you. Always OpenSource. You can also use it with just install the LIMA plugin. I just want to add one more. Basically LIMA is a new feature or new tool set we released in Mirantis OpenStack 7.0. And actually it's based on the OpenSource tools. And backhand actually based on other NOVA and send backhand log database as well. Yeah. Thank you. Yeah. I have a question. Yeah. Since this is a financial application, so what are the performance comparison you are planning to make like before moving to cloud and now, as you said, you are in the middle of the deployment. Sure. So what are the performance indicators you are going to measure in terms of concurrent API or the transactions per minute or the number of devices, you know? Sure. Let me take half of your question and I'll leave another half question to Ms. David. And actually from what I know, actually, Shenzhen Stock Exchange or SSCC already have a small environment. It's around a dozen of nodes based on the OpenStack environment. They are already online. They already take the customer request and they have real customers running on this OpenStack environment already. And I'll leave the second question about the performance. I can translate for you. Okay. So as I mentioned, they already have an OpenStack environment up online and have the real customer running their application on top of that. So right now the customer on the OpenStack cloud is a small business financial institute. And actually David's organization already did some performance comparison and even showed the results to the customer. There's not too much difference. So customer is okay with that. Okay. Thank you. So my question is if SSCC has a special requirement on Marinete's distribution, how do you fulfill their requirement? You want to pull into OpenStack community or you want to pull your own version? Yes. That's a very good question. Actually, after I talked to those so many enterprise customers, actually, SSCC is definitely the pioneer of those customers who want to try OpenStack. And I think I'm proud actually we figured out a way to deliver their request using the Marinete's global OpenStack engineering talent pool along with our Chinese local young engineer talents to combine together to deliver their request. And we actually this morning we even talked to another big player in OpenStack Foundation. We want to, Miss David actually would like to contribute a lot of work they will do in the OpenStack back to the community, especially in the security area actually. Security is a very important concern for financial institute and the system stock changes will do a lot of work in the security to help Marinete's OpenStack or even OpenStack community to enhance the security features and already set some effort not only along with Marinete's even with other platinum number in the OpenStack Foundation we want to contribute back to the OpenStack community. Thank you. How is a financial cloud different from a common cloud? Okay, I think I should be able to provide this. You can add more. Thank you. Actually this is financial, we call it financial industry cloud or we can call it financial community cloud. It's kind of public cloud but it's not open to the common consumer. It's only open to the set of the customer from Sundance Stock Exchange. Those customers mostly are small business financial institute, some small banks maybe brokers, traders, so it's not common consumer, individual consumer. How is the cloud different to satisfy those financial ones? Yeah, how should the cloud be different? There should be some difference in the financial cloud from a common cloud. How is the cloud different to satisfy those different needs? I think actually David mentioned in his presentation secret stable and also the security of the data, right? And also the stable, how stable the cloud it is, it's most important. Actually David even mentioned to me the cost, the price is not his top concern. That's probably somehow a little bit different with the common public cloud. David probably can add more. Yes, I forgot to add that. There are two target customer pools. Right now actually this financial cloud is helping those small financial institute in China because those small financial institute don't have their private cloud. Their IT is very weak, right? So if they want to leverage the benefits of cloud, actually they can use the financial cloud provided by SSCC. And later on once we have this system up and running, we will learn all the lessons, we enhance this system, we think it's ready. Or even talk to those large financial institute to integrate with their private cloud to become somehow the hybrid cloud with them so that they can migrate their workload in the peak time to this public financial cloud. This is actually our goal for this whole project. Thank you. Any more questions? Alright, thank you for coming. Thanks.