 Hello, everyone. Good afternoon. I'm Liu Hongliang. I'm from China, is a stack. And today I'm going to share a user case about Lenovo's user practice on OpenStack. Because of the wide delays that Lenovo's private cloud director Peng Cheng and their project manager Zhiyang couldn't have come to this event. So I will make this presentation for them. And sorry about that. And today we will cover these topics that we are going to introduce a short background about Lenovo and its IT history and also the challenges that Lenovo met in the era of Internet. And we also introduced why Lenovo chose OpenStack to build their private cloud in 2015. And also we will cover how they use OpenStack. And from technology point of view, what kind of workload can run on their private cloud. And also we will cover their future plans on the OpenStack. So here is a short introduction about Lenovo. Lenovo was founded in 1984 and it was very early. It is the very early IT companies in China. And now Lenovo is one of the world's leading personal PC computer companies in the world. And it's a very famous brand in China as well in the world. In 2005, Lenovo purchased IBM's personal computer business. And then Lenovo became an international company. And now it's the largest personal computer company in the world. And also is the first largest smartphone company in the world. And in 2014, Lenovo acquired IBM's Systemics business. So Lenovo also has the enterprise and data center business now. And here is a short history of Lenovo IT systems. Starting from 2005, after the acquisition of IBM's PC business, Lenovo IT also changed them to embrace the international technologies. And as the consumer business market changes rapidly, Lenovo's business strategy evolves as well. As you can see from the table brand strategy to the PC plus strategy in 2013 and now the internet plus strategy from the business point of view. And at the same time, the IT strategy changes as well. Lenovo did a lot of work on the IT management process. For example, version 2, version 3, those IT management process practice for Lenovo's IT management system. And as the internet era comes that lots of the new workloads, like the e-commercial workloads, the big data workloads, such kind of new workloads become more and more popular in their IT system. So there needs also some IT systems to adopt to the workload change. And here is the traditional IT architecture for Lenovo. From the business point of view, there are two models. One is the R model and the other is the T model. And from the IT systems point of view, it's divided to three types. One is the system of record and the system of differentiation. Another is the system of innovation. There are three types of the IT systems. And from the technology point of view, most of the systems are built on top of the commercial products like IBM Power, Windows, SQL Server, IBM Database, DB2, and such kind of commercial products. This is the traditional IT architecture. And what kind of challenges they face using this traditional IT architecture, these are the challenges. As you may know that in the consumer business, lots of changes in PC and mobile market is changed really rapid. And from the internet business model, they need to be very, very, change very fast. They need to be agile. And the application needs to go online very, very quick. And the system needs to be flexible. For example, the peak of the workloads may be very dynamic. And so that requires the IT systems to be agile and to be flexible. But from the traditional IT architecture, it's not easy. For example, it takes several days to one week to deliver a new application. They need to configure the server, the network, the storage and the operating system, the database, the middleware, the application on them. So this kind of installation or deployment of those applications takes a lot of time for them. And that does not meet the requirement for their business changes. And also from the cost point of view that those commercial products are very expensive. And the resource utilization is also low. It's also very low. And from security point of view that the network architecture is hard to isolate applications in data center. And the physical firewall is hard to be flexible for the internet applications. So to deal with those challenges, Lenovo chose to embrace the cloud computing technologies. And lots of options here. So Lenovo did a research on the major cloud computing technologies. So there are lots of technologies that can be chose. There are public cloud, private cloud. And from the technology point of view that there are closed private technology and also there are open source and economic technologies, different choices. The good side of the private technology is that it's consistent. The system is consistent. And the good side for the open source technology is that it avoids the vendor locking. It's more economic. And also Lenovo did a lot of research. And at that time the open source community, there are several open source community like CloudStack, OpenStack, OpenNabla, such kind of open source technologies. And Lenovo's choice was to embrace the open source. And also in 2014, Lenovo became an OpenStack member in 2014. And that made Lenovo's IT department more determined on OpenStack path. And here is the, after lots of testing and POC and the compression of those different technologies, Lenovo chose to use OpenStack. And here is the architecture of the OpenStack system that underlying is the hardware layer. Most of the hardware used is the X86 server. And with the 10 gigabyte network switch for the hardware layer. And for the virtualization layer, Lenovo chose to use KVM for computer virtualization. And they use object storage as, use self as object storage and also self as the block storage. And for network virtualization, Lenovo chose to use the OVS OpenMaze switch. And another point is that there is a Active Directory system in Lenovo. And they need to connect OpenStack with the Active Directory system to input their employees information to the OpenStack Keystone database. And the upper layer is the OpenStack layer. The components are used are Nova, Swift, Cinder, Glance, Neutron, and Keystone. These are all the car projects of OpenStack. And these are the mature projects of OpenStack. So Lenovo chose to use the very mature technology for OpenStack. And above, there is a horizon for monitoring of the systems and also they use heat for authorization. Here is the hardware topology. There are three nodes for the controller nodes. And also the network workloads is run on the controller nodes. These three are the controller nodes, physical nodes. And there is a pool of X86 servers used for computer as well as storage. That means a virtual machine runs on a computer node. And also there is a safe, there is a safe cluster run on the same node. So it's like a computer and the storage running on the same node. So yeah, this is the hardware topology. And from the network point of view, it's divided for the separation of the network flows. So it's divided to four networks. One is the production. So the production is for the workloads, network workloads. And the management network is for the management node to connect to the computer storage node. And the storage network is for the safe cluster to recover the data or to sync the data. And for PSE network is for the installation of the whole cluster. And outside of the open stack cluster, there is a hardware router, hardware router and also a hardware firewall. And there are two network, major network. One is the internet network and another is the internet network. Internet network is public. So a user has access to internet can go public network. And the internet network can only access the network inside the Lenovo. So these are the different two networks. And a cloud user controls which network he use. And these are different IP pools for the different networks. And from the security point of view, there needs to some access control of the network flows. So for example, for example, some of the network protocols are accepted from the direction, from my direction. And some of them is not allowed. It's blocked. And these are the applications running on the open stack cloud. For example, the E2E systems and the E-commercial systems, the education systems and the Model 2 analytics will close. The big data will close. And the co-chat will close. So these are lots of workloads running on the private cloud. And they are planning to move more and more into these systems. The plan is that Lenovo is going to move more and more critical workloads onto the cloud. And from the business model of private cloud that there are major two roles. One is the S-cloud provider. For S-cloud provider, it's mainly the IT department. They are behaving as the S-cloud provider and they operate the cloud. And they provide the S-results to their customers. Their customers are the S-cloud consumer. The consumers are actually the business department inside Lenovo. So the consumer gets the resource from the cloud provider and makes the application and other business stuff on top of the cloud. For example, the consumer can get an AVM and it can install the DVR application on it. And in the future, there is a big plan for a multi-site and multi-region deployment of open stack. It will be built across two cities in Beijing and Wuhan and three data centers. There will be two data centers in Beijing and one in Wuhan. And the workloads can be made great from one data center to another data center. And different workloads, like they can also run a test and pre-production workloads in a different location. And the data can be backuped from one data center to another data center. And also the workloads should be able to migrate from one to another. And also there will be a public connector for private cloud so that the private cloud can also manage a public cloud resource. For example, the AWS or Azure or Ali cloud is a public cloud resource and they can be connected. So when a burst of the workloads happens, the resource can be created from the public cloud. And when the workloads goes down, the results can be destroyed from the public cloud. So this can reduce a lot of cost of the private cloud. Okay, I think that's all for today. So do you have any questions? No, really. It's on AWS or Azure. And one thing I want to mention is that usually the enterprise uses open source technologies in a very conservative way. In China market, the open source technology first used by the Internet companies like Tencent, Ali, such kind of Internet companies, they have lots of engineers to do their own open source. And they get the open source code and they change the code to meet their business requirement. That's a very common model for the China Internet companies. But for the enterprise companies, they use the commercial products to build their IT systems. But there is a turning point that starting from 2015, more and more enterprise customers began to use, to try to use open source technology such as OpenSec. And there are also other projects like big data or containers, such kind of new technologies. So there is a big opportunity for the China market that the enterprise market size is much bigger than the consumer market. And this market, I think there is a very, very large room for the open source technology. They used to use, for example, IBM and VMware, such kind of commercial products. And if they replace all these commercial products with open source, there will be a huge space for the companies like us. And I think it's a very good opportunity for OpenStack as well. Please. Actually, it's not a sink, it's an in-sink. So it's like there is a script that runs on a certain time that it copies the incremental data from one data center to another. And the problem is that the network bandwidth needs to be very stable. And for example, you can trigger a copy of the incremental data at middle net. And maybe it's several times in one day. Yes. It's actually for the CIF, for example, the CIF has three copies for one object. Yeah, this one. So actually the computer has storage on it. So when a write operation for one object, it goes to the private object and then copy to the secondary and the third object. So these all go through the network. And there needs to be a separate network for this kind of data flow. And that's what the storage network is for. It usually is a separate NIC, usually a separate NIC, a 10 gigabit. Yeah. So far it's okay. So the problem is that if you do some large-scale rebalance, the network will be a bottleneck. For example, if you add more nodes to the cluster and there will be a huge rebalance data flow at one very short time. So there needs to be some traffic control to control the rebalance flow. Sometimes we need to reduce the rebalance priority so it can go slower so that the network should not be a bottleneck. But the trade-off is that it takes longer to rebalance. Application server to the database server within your network architecture. Don't you need faster pipe? I'm sorry. So that goes to, I think that goes to the production network. It's also a 10 gigabit network. Yeah. Yeah. No lab. It's tricky that some of the workloads can be migrated. Some of them are not. Usually if the workload is running on a virtualized platform like a Vermeer or Zin, there is a tool to convert the image. And you can export that image and import it to OpenStack after the converter, convert the image. And that takes some time. So actually the business goes down during that time. So there needs to be an outage time. Okay. I think, thank you very much. You have a good evening. Let's go to Beeman.