The authors are he LiNbO, Ding Haiyang and Chen Youkun
- Alibaba cloud container service releases Knight hosting service
Knative is the most popular server less orchestration framework in kubernetes ecology. These resident instances, such as knotic controller and dependent gateway, need to pay additional IAAs cost and operation and maintenance cost, which brings certain threshold to the application of serverless. So we provide managed knave serving in serverless kubernetes. Managed knative is out of the box, and you don’t have to pay any cost for these resident instances. In addition to providing the hosting of controller, we also further reduce the complexity of application serverless through cloud products SLB and burst performance instance retention specification, so that your service can greatly reduce IAAs expenditure during the traffic trough period, and the CPU accumulated during the traffic trough period can be consumed in the peak traffic period, and every penny you pay will not be wasted.
Important progress in upstream
- Graduate Scheduler Component Config API to Beta
In order to promote the use of Kube scheduler configuration API in a wider range` kubescheduler.config . k8s.io is upgraded from version v1alpha2 to v1beta1.
- add “pod-level-single-numa-node policy ” to Topology Manager
A mechanism is provided to ensure that all containers in the pod are bound to the same NUMA node to achieve higher performance.
- KEP for communicating a local registry
It unifies the interaction standard for the local image warehouse used by the local clusters created by King, minicube, etc., so as to reduce the user’s mind.
- Add support for TLS 1.3 ciphers
Add TLS_ AES_ 128_ GCM_ SHA256、TLS_ CHACHA20_ POLY1305_ SHA256、TLS_ AES_ 256_ GCM_ Password support for sha384 type.
- Use maxSkew in score calculation
It is used to solve the problem of uneven distribution of pod between topological domains.
- service controller: only sync LB node pools when relevant fields change
Optimize the service controller synchronization to ensure that the LB node pool is synchronized only when the related fields in the node change.
- Add labels to containers in pod according to annotations
Through the configuration of pod annotation (containers- label.alpha.kubernetes . IO) mark the container.
- Knative: propose another broker based on Kafka implementation
There can be multiple implementations based on broker, as long as it meets the requirements of broker. This proposal removes the ingress and filter components, and the dispatcher implements the filtering.
Open source project recommendation
The tool can query the number of pods that can be deployed in the cluster in real time, so as to help the cluster manager decide whether to increase the machine resources. The implementation principle is to analyze the available resources (including CPU, memory, IO, etc.) in the cluster, and then calculate the number of pods that can be deployed by each node according to the size of the requested resources input by users.
- Yet More Google Compute Cluster Trace Data
Eight years ago, Google opened a batch of runtime state data of Borg, an internal cluster management system. Based on these data, the industry and academia have produced a lot of research results, which also inspired a group of scholars to go deep into this field. At the end of April, Google opened a new batch of cluster data and published the paper “Borg: the next generation”. Interested students must not miss.
Reading recommendations for this week
- 《Choosing a Base Image》
This paper introduces the necessity of selecting the basic image and several common basic images, such as alpine.
- 《Industry Voices — Blaber: Separating vRAN fact from fiction》
This paper focuses on the virtualization and containerization trend of Telecom access network elements.
- 《Certified Kubernetes Administrator Tips and Suggestion》
This paper mainly introduces the preparation suggestions of CKA examination and some problems in the regular examination.
- Flexible and efficient cloud native cluster management experience: managing k8s with k8s
The multi cluster design scheme introduced in this paper, on the basis of many years of operation and maintenance experience of Alibaba group, uses the cloud native architecture to get rid of the dependence on different infrastructure, and realizes runanywhere. Users only need to provide ordinary IAAs facilities, they can enjoy the k8s multi cluster ability which is easy to use, stable and lightweight.
- “Community first OAM visualization platform released! Separation of concerns, user friendliness and low difficulty in getting started
The main content of this paper is a visual implementation based on OAM. At present, the platform is mainly based on the V1 alpha 1 version of OAM, and the core concepts of OAM are abstracted and encapsulated to better meet the user’s habits.
- 《Borg：the Next Generation》
Google published a paper at eurosys 2020 conference. Compared with the data of 2011, the latest Borg system has greatly improved CPU utilization, memory usage, task scheduling efficiency, etc.
- 《Autopilot: workload autoscaling at Google》
This paper introduces Google’s autoscaling system in detail through Borg’s autopilot project, which has a good reference value for VPA and HPA strategy of kubernetes system.
In order to enable more developers to enjoy the dividends brought by serverless, this time, we have gathered 10 + Alibaba serverless technical experts to create the most suitable open course of serverless for developers, so that you can easily embrace the new paradigm of cloud computing – serverless.
Click to view the course free of charge: https://developer.aliyun.com/learning/roadmap/serverless
“Alibaba cloud original concerns micro service, Serverless, container, Service Mesh and other technology areas, focusing on cloud native technology trends, cloud native large-scale landing practice, do the best understanding of the official account of cloud native developers.”