[DISCUSSION] Refactor documents on Fluid's website

[TrafalgarZZZ]

Problems with current documents on Fluid's website:

• Hard to find a way from newbie to expert. Currently there are many concepts introduced in one document, so users, especially the new-comings will find it hard to know where to start and learn Fluid step by step.
  • For example, in Accelerate Data Accessing (via POSIX), we introduce "creating a Dataset", "creating an AlluxioRuntime", "mount the dataset", "speed up data accessing". Such steps cover a full example on how to use Fluid but it'll make a new Fluid user confused about the relationship between them.
• Content redundancy makes high maintenance effort. Most documents demonstrate a full example of how to use the feature. The example usually starts from "creating a Dataset and a Runtime", which makes a lot of redundancy among documents. When API changes, we have to modify each place in every documents.
  • For example, Pod Scheduling Optimization starts from "creating a hbase Dataset" which has nothing to do with the "Scheduling Optimization" topic.
• Lack of examples of XXX feature on XXXRuntime. There are too many combinations about "How to use XXX feature on XXXRuntime", so it's difficult for Fluid to give a tutorial document for each combination. However, users need it!
  • For example, a user can hardly know how to configure a S3-compatible storage as JindoRuntime's UFS, even though he or she has carefully read the document about "using OSS as JindoRuntime's UFS".

So I'm trying to refactor our documents in a more organized and easy-to-understand way. Here is a draft Table of Conetents(TOC) of Fluid's documents based on the TOC here.

## 原文档ToC目录

https://github.com/fluid-cloudnative/fluid/blob/master/docs/zh/TOC.md

## 新版本Fluid ToC目录

- 快速上手（Getting Started）
    - 安装（Installation）
    - 快速体验（Quick Start）
- 介绍（Introduction）
    - Fluid简介（What is Fluid?）
    - 核心概念（Core Concepts）
    - 系统架构（Architecture）
- 教程（Tutorials）
    - 定义数据集（Define Datasets）
        - 配置底层存储数据源（Data Sources）
        - 配置数据集访问模式（Access Modes）
        - 约束数据集调度语义（Scheduling Constraints）
        - 配置数据集放置策略（Placement Strategy）
    - 绑定数据集与缓存引擎（Bind Cache Engine and Dataset）
        - 部署Alluxio（Deploy Alluxio）
        - 部署JuiceFS （Deploy JuiceFS）
        - 部署Jindo（Deploy Jindo）
        - 部署Vineyard（Deploy Vineyard）
    - 访问缓存数据（Access Bound Datasets）
        - CSI挂载模式（Access Data via CSI）
        - Sidecar挂载模式（Access Data via Sidecar）
        - 通过其他接口访问数据（Access Data via Other Interfaces）
            - HDFS接口（HDFS Interface）
    - 动态操作数据集（Operate Dataset）
        - 预热缓存数据（Preload Dataset to Cache）
        - 迁移数据集中数据（Migrate Data from / to Dataset）
        - 以自定义逻辑处理数据（Process Data with Customized Code）
        - 串联多个数据操作（Chain Multiple Data Operations）
        - 设置数据操作自动清理时间（Set TTL for Data Operation）
    - 集成Fluid与自建存储系统（Integrate with On-Premise Storage）
        - 使用ThinRuntime接入任意底层存储（Integrate Arbitrary Storage with ThinRuntime）
        - 加速访问PV存储卷或主机目录（Accelerate Data Accessing for PersistentVolumes and HostPaths）
    - 进阶功能（Advanced Features）
        - 优化业务负载调度（Optimize Scheduling for Workloads）
        - 跨命名空间访问数据集缓存（Access Dataset across Kubernetes Namespaces）
        - FUSE挂载点自愈（Auto-Recover Broken FUSE Mountpoints）
        - 业务运行中添加/移除数据源（Add/Remove Data Sources without Recreating Workloads）
- 运维（Administration）
    - 问题诊断工具（Diagnostic Tool）
    - 数据集缓存扩缩容（Dataset Cache Scaling）
    - 监控（Monitoring）
- 更多配置示例（More Examples）
    - 底层存储数据源配置示例（UFS Mount Examples）
    - 分层存储配置示例（Tiered Store Examples）
    - 元信息策略配置示例（Metadata Policy Examples）
        - 自动元信息同步策略
        - 元信息缓存和超时策略
    - 其他示例（Other Examples）
        - Image Pull Secrets
        - Container Network
        - Options and SharedOptions
        - FUSE Pod Clean Policy
- 开发者指南（Developer Guide）
    - 如何开发（How to Develop）
    - 代码结构和技术原理
    - Fluid Python SDK
    - Fluid Golang SDK
    - Analyze Performance with Pprof
- 参考（Reference）
    - 术语表（Glossary）
    - API参考（API Reference）
    - 版本发布记录（Release Notes）
    - Roadmap

The TOC is made with the following principles:

• One doc for one feature without much redundancy (but should with more explanation).
• Group documents by introduced concepts (e.g. Dataset, Cache Runtime, Mounting, Data Operation, ThinRuntime, ...)
• Concepts are arranged in newbie-to-expert order.
• Put popular configuration examples in "More examples" section. Unlike tutorials, we can just put a Yaml here with little further description.

@RongGu @cheyang @Syspretor @xliuqq Please take a look at it to see if you have any comments on the document structure, titles and other things. Thank you very much!