 Good evening. Yeah, I'm Yuan from Apple. I'm building Kubernetes infrastructure and platform at Apple I have been a very active and code contribute to Kubernetes projects and related projects in the last three or four years this is my 7th cube calm, but the first one in Europe also my 8th talk, but the first Nightingale talk, so it's such a big stage. So Let me Take a quick and photo that I can brag to my daughter Okay, so You can see my name We'll be a proud dad Okay, let's get to back to the topic of my talk and the power where scheduling and in Kubernetes. I I hope yeah, I can to I will bring the attention awareness to it in the next five Minutes So why? So the power outage or the service time the cause the service time is the top one reason in data center and the clusters so now the Kubernetes as being used by more and more applications and Also, the cluster size right is getting larger and larger We have a high risk of this over know that the power supply on the other hand And if you are familiar with the data center power infrastructure We typically also oversubscribed and a power infrastructure to maximize the capital investment Unfortunately and today and the could that is scheduled still only focus on the computer resource like a CPU memory and storage Although and then those are Indirectly correlated to power consumption, but they are not perfect indicate of the power consumption for there are a few reasons One is the request like the port request and worker request don't exactly, right? Represented actual utilization also the power and if you calculate the power consumption It's a correlation to the resource utilization, but they are still idle power other factors can affect it So now this correlation So we argue and we need to bring and integrate this power awareness or make this power as a first citizen as Part of the Kubernetes and we need a bad power management in Kubernetes So what to do here? So the first thing is to prevent the Overload of the power supply. So we need to restrict or limit the aggregator of the power consumption of the workload to this and the power secret at Different level from server to rack enclosure or to the entire data center this power cabin The second is the power awareness and about the power distribution when the schedule to make a decision We are to place the workload. We should make sure right we can better balance this power consumption across the Compute and the power infrastructure to minimize or reduce the risk So how to do it The first thing we need to when they introduce right this and a power Related data from a power topology to the power monitoring data as well as the power model into our the basic and the Kubernetes the Resource or the basic data the secondly that we need a set of the power aware or power efficient the algorithm to support this power capping and the power distribution in particularly We are proposing at two key scheduled plugins. The first one is the field to plug in it will Exclude those nodes or racks infrastructure That already have a very high risk of the overloading the power supply The second one called a scoring in the plug in its preference Right choose the better nodes a better rack to run or load to better balance or reduce this risk to summarize we need and bring this and the power awareness and Make the power of first the citizen in the Kubernetes So including this power to politics and the power monitoring data as well as the scheduling algorithms So also and I used to work on not in a power management data center management in other domain We should leverage and take advantage of the existing work. We are not going to invent a will so this talk of course just a very high level like I said bring this awareness to the community and Another work we can down and even like a previous talk and mohawk mentioned about the copying manage Carbon emissions as well as sustainability So I hope this talk and we inspire the community To work toward and the better power management of sustainability In Kubernetes in the entire communities I also hope I will bring more about these topics in the future and the cube calm Maybe more in the in the deep dive. Okay, that's it I hope everyone enjoying the rest of the conference also one thing in case and you miss it And I highly recommend it go to see the seeds of the tulip flowers. It's amazing and you want and Want to miss it. Thank you very much