 Good afternoon, everyone. I'm from China Fudan University. I'm Zhi Hui. Today our topic is integrated virtualized and containerized architecture for financial cloud big data application. We have a financial team in OpenStack. So today we'll present this topic. Now let's invite our co-operator. Thank you, Professor Lyu. I'm from Intel. My name is Jian Feng. We have many collaborations for this project. As a start, I will introduce the background of our practice for the financial cloud. Currently, that's a collaboration among Fudan University and China Unipay and Intel company. We three part have some collaboration on how to optimize the current China Unipay's financial cloud. Currently, the cloud environment just looks like this. There are several isolated clusters. Maybe some cluster is running virtual machine based on KVM and other isolated cluster where run the Docker containers and also some other cluster running Hadoop or Spark Big Data Cloud. So the question, the problem is, these several cluster are isolated. The hardware resource cannot be shared. So that's the biggest issue. How to resolve it? Our target is here. We need to build one cloud, one unified cloud to let the hardware resource can be shared among all the different kind of Google Cloud including some financial applications and some other big data workloads. Yeah, that's a very draft architecture how we want to do. The left part is our environment current situation. For example, we have isolated virtual machine and cluster based on OpenStack and also some other cluster to run Big Data containerized workload. So these two types of cluster are totally isolated. The workload cannot be talked to each other and the resource cannot be shared also. So the target we want to do, we want to do, we need to set up one unified cluster and this cluster can run the application inside the virtual machine and also meanwhile it can run the Big Data workload inside the containers. So how we do it, professionally, we will continue the topic. Thank you. So you can see in our architecture, You can see in our architecture, we use OpenStack NOVA also, we use these two projects and utilize computer resources. This is placement and then we can locate the VM and container in the same physical resource. So we use this then and Q, Y, R to provide a neutral network to container. We also use placement. We can utilize resource scheduling for both VM and containers. So you can see we use Curve Remote Drivers can create the container network and we can use these drivers to access to Neutron and create the container networks. Also in our architecture, for the Big Data, we use container located, container build the Big Data cluster. Also we use VM build the many finance application. These two, in the located, in the same physical servers, also they can connect to the self or next storage. So we have carried out some experiments to verify our works. You know, in this physical server deployment, in this deployment we have physical server 1, 2, 3. This is around the Hadoop cluster. Also we can change this physical server to the container. We carry out the container Big Data cluster experiment. So you can see these two experiments for the physical server and the performance ratio is 100%. But for the container, we also can get 95%. It's not banned, it's almost good. Also for another experiment, Taylor Sot, we also can, the performance is better than the physical machine. So now we verify that we can use container cluster to run the Big Data. So also we have another test banned. It's means the Fudan University Big Data test banned. In this test banned, we also, in the same physical machine, we can locate the VM, the container. In the future, we also want to have a similar container. It means use the clear container. We can use the Cata project. So in the same physical resource, we can locate the VM and the container. We can build the Big Data cluster. Also we can build many VMs. So this is an example. We call it Sandbox. You can see we can build the Spark clusters. Also for this cluster, we give some detailed information. Also we can monitor the standards of the cluster. Also this is some task running in this cluster. Also this you can see, we have two container clusters. The basic project is based on RAM. But you know the RAM project, just build a single container. But now we can build a container cluster. So in the future, we try to unify this different resource in the same physical resource pool. We can support cluster management and container authorization. This is based on a NOVA. This is based on RAM. But we can unify this different resource in the same physical resource. This is our goal. Also this is our finance team. You know we form the Fudan University, China Union Pay and Inter. We have a financial team in OpenStack. This is our team meeting. We have Vaki and Mayo List. If you want to join in this team, please contact us. Thank you.