GeminiDB Cassandra interface new feature PITR released: supports recovery at any point in time

This article is shared from the Huawei Cloud Community " PITR Release of New GeminiDB Cassandra Interface Features: Supporting Recovery at Any Point in Time ", author: GaussDB database.

technical background

When a series of failure scenarios such as data corruption, data loss, and accidental deletion of data occur in the business, the database often needs to be restored to a certain moment before the failure occurred, and the smaller the granularity of the recovery, the better. When traditional databases use periodic backups for data recovery, they can only be restored to the backup time point. Due to limited backup points, more fine-grained point-in-time recovery cannot be performed, which greatly affects the user experience.

Huawei Cloud Database GeminiDB is a cloud-native NoSQL database based on Huawei's self-developed computing and storage separation architecture and compatible with the Cassandra ecosystem. After the comprehensive backup cost, recovery timeliness and granularity are fully verified, the GeminiDB Cassandra interface launches a new feature PITR (Point-In-Time-Recover) to support recovery at any point in time. You can freely choose the time point that needs to be restored, thereby protecting users Data consistency reliability and security!

cke_138.png

Implementation principle

Overall backup cycle

A complete backup cycle = 1 full data backup + regularly run incremental log backup + interspersed differential backup (the frequency is adjusted according to the business read-write model).

cke_139.png

Full backup

PITR supports the realization of recovery performance at any point in time, partly relying on a full data backup.

Full data backup provides the data basis for subsequent incremental log backup. When performing backup, the GeminiDB Cassandra interface will take a second-level snapshot of the data and make full use of each node in the cluster to perform data segmentation and uploading, thus achieving no business awareness. It does not affect the performance of the instance.

cke_140.png

Incremental log backup

PITR supports the realization of recovery performance at any point in time, and the other part is inseparable from incremental log backup.

Incremental log backup is a backup method based on log files WAL (Write-ahead log). Compared with traditional incremental data backup, it has the characteristics of light backup task volume, fast upload speed, and space saving. The GeminiDB Cassandra interface will regularly back up archived WAL and clean and recycle it in a timely manner after uploading.

cke_141.png

differential backup

PITR supports the realization of recovery performance at any point in time, and is also related to differential backup.

Log playback will be performed when the PITR instance is restored. When the full backup cycle is set to be long and the business write volume is large, it will cause too many logs during playback, causing the playback time to increase. Differential backup can alleviate and avoid this problem. Differential backup is essentially a data backup that compares and uploads data based on the last full backup or differential backup, effectively reducing the amount of log playback and improving the recovery rate.

cke_142.png

Recovery process

  1. Download full backup data to new instance
  2. Download the incremental WAL file to the new instance
  3. Open the database in recovery mode for WAL log playback
  4. The process of restoring to a new instance ends

cke_143.png

Summarize

PITR, a new feature of the GeminiDB Cassandra interface of Huawei Cloud Database, realizes recovery at any point in time, filling the shortcomings of traditional NoSQL engines that can only restore at a specified point in time. The recovery accuracy is accurate to any point in time, which greatly improves the user experience and provides data Provide better security and reliability.

appendix

Click to follow and learn about Huawei Cloud’s new technologies as soon as possible~

IntelliJ IDEA 2023.3 & JetBrains Family Bucket annual major version update new concept "defensive programming": make yourself a stable job GitHub.com runs more than 1,200 MySQL hosts, how to seamlessly upgrade to 8.0? Stephen Chow's Web3 team will launch an independent App next month. Will Firefox be eliminated? Visual Studio Code 1.85 released, floating window US CISA recommends abandoning C/C++ to eliminate memory security vulnerabilities Yu Chengdong: Huawei will launch disruptive products next year and rewrite industry history TIOBE December: C# is expected to become the programming language of the year A paper written by Lei Jun 30 years ago : "Principle and Design of Computer Virus Determination Expert System"
{{o.name}}
{{m.name}}

Je suppose que tu aimes

Origine my.oschina.net/u/4526289/blog/10320514
conseillé
Classement