 Thank you. We have full, we have a few people so I'm not so nervous first. One minute, everybody. I'm Zhao Huan Yu from China, from the State Grid Corporation of China. I have 10 years about IT experience and I'm a database administration team leader and an OpenStack maintenance team leader. We can see there are two pictures on the left side. The upper one is just after graduation before work. I'm slim, maybe I have some weight, but after 10 years I've become a fat old uncle, lost almost all my hair. Why? Because we have to work in SGCC IT department. We must fix bugs, fix on the symptoms, so many works, just sit down. So no time to do more exercise, no time to eat gym. So I became a fighter fighter. But with OpenStack in our company one year, I become slim again because I have more time in gym. So OpenStack do good for me. Okay, let's move to our slide. We have four parts in this slide. The first is about SGCC and its IT architecture. SGCC is I think is most new to all of you because it's China's largest electricity company in the world. It's the 7th largest company in top 500. So we serve about 26 states in China. Most of the red area is served by us. Before OpenStack, before Mootoo open source, we mainly use commercial products in our systems, both soft wires and hard wires. So we can see we are almost locked by the vendor. For this, Mootoo open source, a lot of things are moving to open source. We build our own self controlled private cloud because we have 26 states. So it's a big cluster for computers. So we must have our own private cloud. Okay, this is about our company. Power supply is key to protect people's new hold and economic development for our society because power is most important for us. Without power, we can't do anything. If we do not have any power in our home, in our office, then nothing is done. So power is very important. So the power supply must be safe and reliable. IT systems are the key points for the power supply to work safely and reliable because everything in my office, in my company is working on the IT systems, emails, website, financial systems, production system, ERP, CRM, planning systems. A lot of systems are working on the information systems. So if the information system is done, you can imagine it. This slide is talking about the traditional architecture of our IT systems. Many of our database and storage is the most important part in our IT systems. We have more than 100 information systems in the whole company, in the whole corporation. And we have 500 terabytes in each state. We have 26 states, but we have 500 in each and 1,000 hosts in each state. Most of the hosts and the survivors are locked by the vendor. So we are not comfortable. We must move to open source. Okay, then we should move to the open source. Then we have some purpose on the changes. So this slide is talking about our IT architecture. The most important thing for us is reliable. The systems must be stable, reliable, and safe because IT information systems are most important. Another thing is we must unlock by any vendor. The last thing is costs. We should control the IT investments, IT costs. We must save money. Okay, this slide is talking about how to unlock from the vendor. For this slide, it's a typical architecture of a database system. We have a mini computer and a switch, fiber switch, and storage. It's a traditional database system. So we must change. We must unlock from IBM X. We must change from EMC storage. More and more technologies on the new hardware, software appear in these five years. We have so many things to play. So it's the key reason we can change. Another thing is the old infrastructure is moving to infrastructure as a service. I mean, everything is moved to cloud. Another thing, the last thing is about commercial software and hardware. If you want to upgrade from the before one, you must pay more because the commercial software and hardware, they want to scale up, not scale out, but they want to scale out. They want to have more servers. Another thing that can happen, if a new system is started for test deployment and online for working at the first time, and with more users, we need more RAM, more CPU, more storage, then we must attend the systems. For the old method, we must scale up. It's not good for me, for us, I think. We want to make some big changes. So we use X86 to take instead of units. We want quantity, not quality. Quality do not mean that with bad quality. I mean, mini-computer is always stable than the PC server, but if we have more PC servers, it's more stable than the one-hour-two mini-computer. So it's quantity for quality. Another thing is we change the commercial software to open source. In the latest two years, we have more projects with open source software. You know how it comes in? Okay, that's coming to the open source. It's our headache that the commercial software, we must pay for lessons. We must pay for the service. Everything in their books, you cannot control the data, control everything inside the books. So it's commercial software. But we want to control everything. We want to hold our data in our own. So it's not good ideas for commercial software such as VMware. I think VMware is not here. VMware is a very expensive lessons, and Oracle is our big problem, because our database is Oracle 10G, 11G, and middleware is also WebLogic 10G, 12C. It's both from Oracle. If we combine from Oracle Xdata, Xlogic, it's all from Oracle. So you cannot control. They just ask for money. We ask for service. They ask for money. So our money is done. But our financial staff said we do not have money, so we must control the investment on IT systems. So from 2014, my team started because I'm the database administration leader. Our team started to remove IBM X and EMAC storage. I mean the commercial storage to open source. So we started in March and in November. For all the database systems, we removed IBM X and commercial storage just eight months. We removed them all. Because Oracle, we can't remove because it's depending on the development. So we moved the infrastructure to our architecture. And 2014 December, we began to know OpenStack. With almost 10 months' tests on deployment and selection of OpenStack vendor, our OpenStack partner, we chose EasyStack as our OpenStack partner. We started to implement the OpenStack environments. Until this February, we have 200 physical nodes in one state, just one state for the test for the lab. And we have more than 1,000 virtual machines for our IT systems. It's not too long. I think it's about a half a year. But we're planning to do more. We have more systems waiting for the OpenStack. Now we're also working for the big data cloud. As the X, I mean network storage, everything we want to define by the software. Okay, let's move to the next part. It's OpenStack now. From the last slide, I said that I chose EasyStack as our partner of OpenStack. That's why we chose it. Because EasyStack is the leader OpenStack provider in Asia, in our country. And they focus on the professional enterprise software, enterprise service. It's very important. Some public cloud, we have also tested some public cloud provider. But they have different thinking, different ways. So it's not a good idea to choose them. Because we want to keep our data under our control. So it's better to choose a private cloud for this enterprise. And if our partner, they must provide all the public and private API. That means it's 100% open and locked by any vendor, our partner. And the OpenStack systems must be reliable. You should understand enterprise better than you can do better. If you are new to enterprise, it's totally different from the Internet company. Because we are a public service company. And it's best performance and best values. Okay, I'd like advertisement for them. Now it's our current OpenStack environments. We have more than 200 nodes in this one. I mean, in this February, we have 200 nodes. We plan to have 1,000 nodes here. And in the 200 nodes, we have five controllers. And more than 130 commutes nodes. More than 60 storage nodes. This is our logical map for the OpenStack. This one is a standard rack. We have 10 servers in one rack. And mixed with storage nodes and commutes nodes. We have three switch in one rack. And a two-call switch for the up-level. It's our logic map. This slide is talking about the network. We have five totally different networks in our OpenStack environment. We have, okay, sorry. There's some Chinese character. It's just from my slide in China. We have internal network. It communicates between VMs. It means VM is transferred from this host to another. We have separate network for them. And we have external network. It means provide service for the user. We have storage network at around the site only. They should pay more attention on the site, on the storage network. Because they have some assailants in this network. I will talk later. And we also have the volume deployment network and management network. But it's just used for once or several times, not used very often. So we just, we put them together, run in a gigabyte switch, network switch. This means admin node and commutes node and storage node. Okay. OpenStack is always flexible and customized. That means if in our company, the user, the user, they want to some new service, some new request. We must development for them. Because we should service for them for the users. For commercial software, you can't develop some service. But you just can't, you just development from the open API. So internal things, you know nothing about it. But with OpenStack, every API is open. So we can have new request and development. And then we have new service, new module. Then we get a better performance. Because we, OpenStack can work with Docker for container. For monitor and unstable for deployment and management. So it's quite easy for us to development the other part. We totally move to the open source. Okay. Then it's the map of our IT system. The hardware, we only need servers. White box servers. White box switch. And the fiber cable and the rack. It's our hardware. And then the operating system of cloud. We choose OpenStack. And the other part, we choose Xavius. Unseable. We have test on the deployment. So many open source tools for our systems. It's almost the automation operation. So we have nothing to do. Most of us, we have more time to play. So for us, after OpenStack, we have more time on holidays. Okay. This is both of the future jobs in our company with OpenStack. First, we should enlarge our OpenStack cluster. We should expand the OpenStack cluster planning implement more than 1,000 nodes in each state. I mean that maybe in the future, in the future five years, in the next five years, we have 50,000 nodes in the whole corporation company. Maybe it's the largest production of OpenStack environment in the electricity industry in the world. Because we are already the first one in the world in the power supply. So if the OpenStack cluster is so big, it's becoming the biggest one. Biggest production environment. And we plan to upgrade to a new version because now our OpenStack is at-house version. We plan to move to a new unstable version of OpenStack. And also, I'm the database administration. I just focus for my work. So I have paid more attention to the storage. We plan to do some self on storage. We want to implement the database cloud with self. And we plan to have some QoS on the storage network. Because if the storage network is flashing, then nothing can run. So it costs almost all the bandwidth. Okay, a little fast, I think. The last one is the suggestion. This is just a suggestion from my team. So it's just a suggestion. The first one is the technical suggestions. We must choose the stable version of OpenStack, the stable version of OpenSouth. Because OpenSouth projects have so many bugs. We must care about it. And we must have good quality on hardware. And SDN is better than the traditional network. We are working from transfer from the traditional network to SDN by switch, by part, by zone, by zone. And the SSD hard disk will come popular in the enterprise area, I think. We use SSD more and more for these days. For non-technical issues, we should improve our IT skills. We should improve our DevOps. And also we should rebuild the structure of IT department because we do not need mini-computer administration anymore. It's totally removed from our systems. So we should rebuild our IT department structure. And the last thing is that we have to change our thinking and thinking on our method of how to operation and maintenance the cloud, in-cloud. I mean, it's totally different from the traditional IT architecture. So my part is finished. It's quite quick. Okay. And thank you. It's finished. Thank you. Any questions? No questions? Okay. Thank you. Thank you very much. I would like to ask two questions. Okay. Last slide. You told that IT skill improvement is very, very needed. So what skill is most needed for IT skills? Yes. Next one. Rebuild software of IT department. How change your department before and after? Okay. Okay. I know. For the first question, we do not have mini-computer in our system. So the IT skills, we have more PC servers. So the skills, the knowledge is transferred from the mini-computer to PC server. Because it's totally different. We have unique systems. So unique system administrator, the high level in the before days. But the PC server with Linux is very popular. It's easy to understand. But we must have so many Linux administrations. So it changes our IT department structure, I think. Both questions are answered. To Linux. Yeah. Yeah. Yeah. Because Unix is quite expensive. For this, there are few and few people to use Unix. So yeah. Yeah. I have some, because it's not my job. I just maintain the database system and open stack. If we need some more programming jobs, we move to our partner. Okay. Yeah. Okay. Thank you. So I have a question about the size of your cloud. Okay. And in one of your slides, you mentioned you have five controllers, 60 plus storage nodes. Are those nodes part of your 200 nodes cloud? Yes. Yes. Both the inside. Okay. So totally it's 200. So because I'm more interested in the compute nodes to storage ratio. Racial. Yeah. I think with the slide, you can see it. So you probably have over 100 compute nodes? Yes. Yes. Oh, no, no, no, no. I can see it. Sorry. That's okay. I can do the math. I mean, if you say that the controllers and storage, they're part of the 200 nodes. Yes. You can see the picture. Okay. I will move on. Yes. You can see that we have five controllers, more than 130 compute nodes and 60 storage nodes. Okay. Thank you. So the other question is about deployment because anyone working on the open stack nodes is hard to set up. Yeah. Do you use some of your in-house develop tools or use some standard deployment tools or? Yeah. Part of it belongs to our partner, but I also can answer that they have a roller. A tool means a roller. With deployment, it's an open stack very easily with the roller. It's a tool, a software. It's part of the easy stack, right? Yeah, yeah, yeah, yeah. We implement about 50 nodes just within one day with the roller. It's very fast and it's stable. Okay. Thank you. Welcome. Okay. Then it's finished. Thank you. Again.