This repository has been archived by the owner on Nov 28, 2023. It is now read-only.
index.md translated #1
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
| [](http://59.111.93.165:8080/job/curve_multijob/lastBuild) | ||
| [](http://59.111.93.165:8080/job/curve_multijob/HTML_20Report/) | ||
|
|
||
| CURVE is a distributed storage system designed and developed independently by NetEase, featured with high performance, high availability, high reliability and well expansibility, and it can serve as the basis for storage systems designed for different scenario (e.g. block storage, object storage and cloud database). |
There was a problem hiding this comment.
分布式存储系统的扩展性一般用 Scalability （expansibility）
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
| [](http://59.111.93.165:8080/job/curve_multijob/lastBuild) | ||
| [](http://59.111.93.165:8080/job/curve_multijob/HTML_20Report/) | ||
|
|
||
| CURVE is a distributed storage system designed and developed independently by NetEase, featured with high performance, high availability, high reliability and well expansibility, and it can serve as the basis for storage systems designed for different scenario (e.g. block storage, object storage and cloud database). |
There was a problem hiding this comment.
// it can serve as the basis for storage systems designed for different scenario(......)
it can serve as a storage base for different scenario(......)
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
| CURVE is a distributed storage system designed and developed independently by NetEase, featured with high performance, high availability, high reliability and well expansibility, and it can serve as the basis for storage systems designed for different scenario (e.g. block storage, object storage and cloud database). During the development of CURVE we sticked to three principles: | ||
| - Follow the trend of storage device and infrastructure nowadays to build top-level storage products that coordinate hardware and software perfectly. | ||
| - 'Simple can be harder that complex'. As long as we can solve the problem well, we'll choose the most simple way. | ||
| - Embrace open source components. We use excellent open source components assessed and tested by us to avoid 'reinventing the wheel'. | ||
|
|
||
| So far, we have implemented a high performance block storage system, which supports snapshot, clone and recovery, and can be attached on QEMU virtual machine or physical machine (by curve-nbd). CURVE has been served as an elastic block storage service inside NetEase for a certain time, during which high performance and reliability have shown. |
There was a problem hiding this comment.
(by curve-nbd) 这里为啥要加括号
There was a problem hiding this comment.
这里是当时盼总的修改，我觉得准确性可能高一些就留了原文。直接用physical NBD是否准确？
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
| CURVE is a distributed storage system designed and developed independently by NetEase, featured with high performance, high availability, high reliability and well expansibility, and it can serve as the basis for storage systems designed for different scenario (e.g. block storage, object storage and cloud database). During the development of CURVE we sticked to three principles: | ||
| - Follow the trend of storage device and infrastructure nowadays to build top-level storage products that coordinate hardware and software perfectly. | ||
| - 'Simple can be harder that complex'. As long as we can solve the problem well, we'll choose the most simple way. | ||
| - Embrace open source components. We use excellent open source components assessed and tested by us to avoid 'reinventing the wheel'. | ||
|
|
||
| So far, we have implemented a high performance block storage system, which supports snapshot, clone and recovery, and can be attached on QEMU virtual machine or physical machine (by curve-nbd). CURVE has been served as an elastic block storage service inside NetEase for a certain time, during which high performance and reliability have shown. |
There was a problem hiding this comment.
CURVE has been served as an elastic block storage service inside NetEase for a certain time
elastic的修饰是什么意思？
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
|
|
||
| #### MDS | ||
|
|
||
| MDS serve as the center node of CURVE, responsible for managing metadata and scheduling. Metadata includes system topology information, namespace of the file system (e.g. tree-based directory, directory metadata, etc.) and copyset placement data. Scheduling is to monitor the status of the cluster and coordinate their running, including detecting on/offline status and collecting workload data of chunkservers, balancing load of clusters and recovering data from failures. MDS support high availability by leader election of etcd clusters. Data between leader-MDS and follower-MDS will not be synchronized, instead, follower-MDS will restart the service by reloading data from etcd. |
There was a problem hiding this comment.
//1. Metadata includes system topology information, namespace of the file system
system可以不用
ilixiaocui
approved these changes
Aug 10, 2020
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
| - [Client Python API](docs/cn/curve-client-python-api.md) | ||
| - Application based on CURVE | ||
| - [Work with k8s](docs/cn/k8s_csi_interface.md) | ||
| Client provides near-POSIX interface for applications, and implements metadata and file data operations by the interactions with MDS and chunkservers, also, it splits the I/O, operates specified QoS control on IOPS and bandwith. Hot upgrade is suported by Client, its version can be updated without user's awareness. |
There was a problem hiding this comment.
// Hot upgrade is suported by Client, its version can be updated without user's awareness.
Hot upgrade is suported. Update of client has no effect on running service.
ilixiaocui
reviewed
Aug 10, 2020
index_en.md
Outdated
|
|
||
| Want to try on it? Take it easy! We'll help you step by step, but make sure you've read this [Special Statement](docs/cn/deploy.md#%E7%89%B9%E5%88%AB%E8%AF%B4%E6%98%8E) before you start. | ||
| The snapshot and clone service of CURVE block storage system is provided by a seperated subsystem, implemented by a component called SnapShotCloneServer. Incremental and full snapshot are both supported, and for saving spaces, snapshot that users created will be uploaded to S3 clusters. |
There was a problem hiding this comment.
SnapShotCloneServer is independent of core services. Curve support Incremental and full snapshot . For saving spaces, snapshot of files will be uploaded to S3 clusters.
ilixiaocui
reviewed
Aug 11, 2020
index_en.md
Outdated
|
|
||
| [Deploy on multiple machines](docs/cn/deploy.md#%E5%A4%9A%E6%9C%BA%E9%83%A8%E7%BD%B2) | ||
| High performance is one of the core features of CURVE, and also our motivation of starting this project. For RPC implementation we used [brpc](https://github.com/apache/incubator-brpc), an open-source high performance and low latency industrial-grade RPC framework. As for consensus of replicas, we implemented by [braft](https://github.com/baidu/braft), which is an open-source quorum-based Raft algorithm implementation. In the aspect of protocol, quorum mechanism can do better than strong replica consistancy model in latency. In our implementation, we optimized the snapshot of braft by using chunkfilepool (specify a certain portion of space for chunk during the initialization of the cluster) on the replicated state machine of Raft to achieve zero write amplification. Also, CURVE implements a finer grain of data address hashing on chunks to achieve read/write splitting and reduce I/O collision, which improve I/O performance furthermore. |
There was a problem hiding this comment.
// 1. and also our motivation of starting this project
这句可以不要
// 2. we optimized the snapshot of braft by using chunkfilepool (specify a certain portion of space for chunk during the initialization of the cluster) on the replicated state machine of Raft to achieve zero write amplification.
可能理解有点错误，是两个方面，raft的快照 和 raft的状态机
CURVE implemented a lightweight snapshot of braft. （实现了braft的轻量级快照）
CURVE also use chunkfilepool (specify a certain portion of space for chunk during the initialization of the cluster) on apply to state machine to achieve zero write amplification. （curve在raft中状态机的实现就是实际的写chunk, curve使用chunkfilepool，让实际写chunk的时候写放大为0）
ilixiaocui
force-pushed
the
master
branch
7 times, most recently
from
2d104c9 to
5b58652
Compare
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Translated version of index.md added.