Cloud Native Computing Foundation KubeCon + CloudNativeCon Europe 2017 (Berlin), 6 Apr 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keynote: Cloud Native in the Enterprise: Production, Best Practice and Innovation - Dr. Ying Xiong

Description

Keynote: Cloud Native in the Enterprise: Production, Best Practice and Innovation - Dr. Ying Xiong, Chief Architect of Cloud Computing, Huawei Technologies

Moving enterprise to cloud native is hard and challenging. In this brief talk, we would like to share our Kubernetes production use cases, and the values created for Huawei enterprise. The talk also discusses some best practices in building Kubernetes based PaaS platform for production, as well as innovation to solve real customer problems.

About Ying Xiong
Dr. Ying Xiong is Chief Architect of Cloud Platform at Huawei Technologies. His current responsibilities include the architecture vision, strategy and design of cloud platform as-a-service (PaaS). Ying brings 20+ years of experience in architecture and design of cloud, ecommerce and enterprise IT systems. Prior to Huawei, Ying worked about 8 years at Microsoft as principal architect and development manager in Azure, and SQL Azure cloud platforms. He also spent over 10 years at AT&T as principal enterprise technical architect and development manager. He has a Ph.D. and B.S degree in Computer Science.
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.

A

Good morning, everyone and again my name- is Aisha I'm chief artifact, our platform, our hallway, it's good to be here at the same safe conference and uh personally, a happy new off in the same safe since the beginning back to jawline and 2015. So in the last one and a half years, I haven't seen any enormous growth of the same say for both from technology perspective and also from the organizational perspective. So it really becomes a home for the cloud native scenes. So we are proud to be part of it. Actually. Looking for the slice.

A

All out million I was looking for this man. Excuse me.

A

Got it thank you, I got it.

A

Okay, got it thanks so uh back to last year's conference at Seattle right, reduce the old cloud strategy, they're happy net of our way at the center of the strategy is the cloud native platforms that across unified across the company- and this is, is a platform where cold fusion stage is a path product based on the kubernetes.

A

This platform now is supporting many internal and external customers, the help indent for move to the cloud native architecture. So today I want to share with you some of the use case scenarios we have. One of the scenarios is over only to know IITs I want to share with you our experience and of the best practice.

A

We learn during the process of moving to the cloud native architecture, so a need to be a background where IT has a hundred modern, a hundred applications that deploy around the globe cross a data centers and those a hundred plus applications is running close to 1 million virtual machines weighs hundred seventy thousand consistent users, one meaning what emotion? That's a lot of erasers for a hundred, so one of the big problems they have.

A

You know where Amanda other pinpoints is that resource utilization problems they just very quickly ran out, and we end that need for new applications for new instance so and also the resource utilization spore way. M is really low, so they started looking at the cloud native architecture back in 2015, they start to redesign some of the apps they have with the micro services architecture. They start to internalize some of the applications, but end of last year, 2016.

A

They are able to using the platform to managing more than four thousand knows, they're running twenty thousand containers in the run in the production environment. So they can't Saugus idea. They say that the goal for this year is to running a happy thousand containers in the production environment. Well, this is really a big target for them anyway, but they, this cloud that he plant on the hill, tend to run more apps with the same amount of resources.

A

In some scenarios they are able to reduce the number of aims from Henry T 2002 20s for the same same amount of workload.

A

Additionally, the table sometimes table the Association for VM and also they're able to improve the global deployment time to 80,000 us from one week down to the minutes. You know this is a real value that cloud, ladies, can bring to the enterprise so along the way from POC to production within some valuable nations and with some of the best practice at the first, you have to choose the rest out of application to start with your cloud native journey, some applications just too difficult to redesign with the macro service architecture. You don't want that.

A

You don't want to start with two certifications. Secondly, you need a strategy for managing your containers at the wrong time. You need to ask the question: do I need to access my container at the wrong time? Do I need persistent storage, that from container data, do I need around multiple processor per container, which you can run? You can run multiple process upon container to make contents to need to communicate with back in the system they're still running in the wave or physical machines you need, because you need to set up in the network for them.

A

So these questions they help, you define the right strategy and the right process for you to manage the containers at the wrong time. Additionally, in most cases you were wrong. Are multiple cluster in your cluster, in your production environment, so having a global cluster management strategy is critical to your solid enterprise. Well, thanks to the kubernetes Federation project, there are makes things much easier to manage across the multiple clusters, but there are still many things to be done. The white zone we did is we build a point-to-point algorithm to synchronize the continent image across eight-thousanders.

A

This really helped us to reduce the deployment times form in the awake down to the minute.

A

Similarly, for the network, containers or container network, you need to understand or know your network requirements. You need to figure out the network performance for your containers, the networking latency for your containers and those eventually decide your network solutions for your container applications. For us. We have been working on a solution called ICANN, which is supporting many different network schemes. We support, you know our lay underlay architecture. We support neo to nursery label for networking. We support villains and net Anasuya.

A

Just this morning, I receive an email that talking about the requirements from customers and they want a multiple IP address per container. Those are requirement you need to understand for you to define your strategy, how to manage the network for your container application and it's critical to to your enterprise cloud in your enterprise. I.

A

Like this feature a feature, we call hybrid application, orchestrations now wakes Larry to explain the architecture of kubernetes by introduced the process. Controller shear in the diagrams now I have a one platform can manage container based under non container based applications within the same clusters we found out. This is very useful for our customers.

A

For some of existing applications, hoping in the move from the naked say, applications to the cloud data for architecture- and this feature also help our own IT- to reduce 95% of manual work which needed for them to manage the non container applications is a cool feature.

A

Last panelist is you have to somebody file, your storage management I. Don't need emphasize how important this is for your enterprise. Now for us, we are working with the company in our MC hatch, it Fujitsu Hitachi to allow the platform, the dynamic, provisioning, the storage with standard API. We are hoping that we, this work onwards, raises a side, the container storage interface, that's community, working on I.

A

Think, that's all for me. Thank you very much for your time. I hope you enjoyed the recipe more.