当前位置:网站首页>Cloud native (32) | kubernetes introduction to platform storage system
Cloud native (32) | kubernetes introduction to platform storage system
2022-07-06 23:12:00 【Lansonli】
List of articles
Kubernetes Introduction to platform storage system
Kubernetes Introduction to platform storage system
The storage system is a professional whole system , We only use common methods to build , The specific optimization needs to be adjusted with reference to the company's own products
One 、Ceph
Official website :https://ceph.io/
1、 Basic concepts
Ceph There can be
Ceph Object storage : Key value storage , The interface is simple GET,PUT,DEL etc. . Such as seven cattle , Alibaba cloud oss etc.
Ceph Block device :AWS Of EBS, Qingyun's cloud disk drive and Alibaba cloud's Pangu system , also Ceph Of RBD(RBD yes Ceph Block storage oriented interface )
Ceph file system : It has richer interfaces than block storage , Need to consider the directory 、 File attributes, etc , Implementing a file store that supports parallelization should be the most difficult .
One Ceph The storage cluster needs
At least one Ceph The monitor 、Ceph Manager 、Ceph OSD( Object storage daemons )
Need to run Ceph File system client , You need to deploy Ceph Metadata Server.
Monitors: Ceph Monitor (
ceph-mon
) The monitor : Maintain cluster state informationMaintain the mapping of cluster state , Including monitor mapping , Manager mapping ,OSD mapping ,MDS Mapping and CRUSH mapping .
These mappings are Ceph Critical cluster state necessary for daemons to coordinate with each other .
The monitor is also responsible for managing authentication between the daemon and the client .
Typically, at least three monitors are required to achieve redundancy and high availability .
Managers: Ceph Manager Daemon (
ceph-mgr
) : Responsible for tracking runtime metrics and Ceph The current state of the clusterCeph Manager Daemon (ceph-mgr) Responsible for tracking runtime metrics and Ceph The current state of the cluster
Including storage utilization , Current performance indicators and system load .
Ceph Manager Daemons also host based on python Module , To manage and disclose Ceph Cluster information , Including based on Web Of Ceph Dashboard and REST API.
Usually , At least two managers are required to achieve high availability .
Ceph OSDs: Ceph OSD ( Object storage daemons ,
ceph-osd
) 【 Store the data 】By checking other Ceph OSD The daemon's heartbeat to store data , Dealing with data replication , recovery , Rebalance , And to Ceph Monitors and managers provide some monitoring information .
Usually at least 3 individual Ceph OSD To achieve redundancy and high availability .
MDSs: Ceph Metadata Server (MDS,
ceph-mds
ceph metadata server )Storage can represent Ceph File System Metadata ( Such as :Ceph Block device and Ceph Object storage is not used MDS).
Ceph The metadata server allows POSIX File system users execute basic commands ( Such as ls,find etc. ), Not to Ceph Storage clusters are a huge burden
Two 、Rook
1、 Basic concepts
Rook It is a storage orchestration tool for cloud native platforms
Rook Here's how it works :
Rook The structure is as follows
RGW: by Restapi Gateway
2、operator What is it?
k8s in operator+CRD(CustomResourceDefinitions【k8s Custom resource type 】), It can quickly help us deploy some stateful application clusters , Such as redis,mysql,Zookeeper etc. .
Rook Of operator It is our k8s A parser that interacts between a cluster and a storage cluster
CRD:CustomResourceDefinitions ( Custom resources ); Such as :Itdachang
operator: This can handle custom resource types
- Blog home page :https://lansonli.blog.csdn.net
- Welcome to thumb up Collection Leaving a message. Please correct any mistakes !
- This paper is written by Lansonli original , First appeared in CSDN Blog
- When you stop to rest, don't forget that others are still running , I hope you will seize the time to learn , Go all out for a better life
边栏推荐
- asp读取oracle数据库问题
- OpenSSL:适用TLS与SSL协议的全功能工具包,通用加密库
- Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medi
- DockerMySQL无法被宿主机访问的问题解决
- Machine test question 1
- Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medi
- Method of canceling automatic watermarking of uploaded pictures by CSDN
- Is "applet container technology" a gimmick or a new outlet?
- ACL 2022 | 序列标注的小样本NER:融合标签语义的双塔BERT模型
- Let's see through the network i/o model from beginning to end
猜你喜欢
每人每年最高500万经费!选人不选项目,专注基础科研,科学家主导腾讯出资的「新基石」启动申报...
Flutter life cycle
How to confirm the storage mode of the current system by program?
Dayu200 experience officer runs the intelligent drying system page based on arkui ETS on dayu200
CSDN 上传图片取消自动加水印的方法
European Bioinformatics Institute 2021 highlights report released: nearly 1million proteins have been predicted by alphafold
ICLR 2022 | 基于对抗自注意力机制的预训练语言模型
云原生(三十二) | Kubernetes篇之平台存储系统介绍
Aardio - Method of batch processing attributes and callback functions when encapsulating Libraries
Docker starts MySQL and -emysql_ ROOT_ Password = my secret PW problem solving
随机推荐
(shuttle) navigation return interception: willpopscope
How to use flexible arrays?
让 Rust 库更优美的几个建议!你学会了吗?
QT信号和槽
DockerMySQL无法被宿主机访问的问题解决
浅谈网络安全之文件上传
Children's pajamas (Australia) as/nzs 1249:2014 handling process
(DART) usage supplement
CSDN 上传图片取消自动加水印的方法
ACL 2022 | 序列标注的小样本NER:融合标签语义的双塔BERT模型
How big is the empty structure?
Interview question: AOF rewriting mechanism, redis interview must ask!!!
安全保护能力是什么意思?等保不同级别保护能力分别是怎样?
Detailed explanation of ThreadLocal
How to confirm the storage mode of the current system by program?
Word2vec (skip gram and cbow) - pytorch
AcWing 4299. Delete point
Improving Multimodal Accuracy Through Modality Pre-training and Attention
POJ 1094 sorting it all out
Let's see through the network i/o model from beginning to end