As one of the basic software, the domestic database has always been a relatively weak link in the information industry.
However, as the development of digitalization enters the fast lane, the domestic database ecology ushers in vigorous development, and highly innovative products continue to emerge, which promotes the blossoming of database applications everywhere. Among them, openGauss is the best.
At the openGauss Developer Day 2023 conference held recently, Huang Kaiyao, chief architect of the openGauss open source database and member of the community technical committee, shared with you a number of architectural innovations in the latest version 5.0 of openGauss, and gave an in-depth interpretation of its new features. After the meeting, we also had the honor to interview Mr. Huang Kaiyao, and had a more in-depth chat with him about openGauss and some of his views and insights on the domestic database field.
twin engine powered
What is the strength of openGauss 5.0?
Looking back on the past, behind the iteration of database technology is the change of data model, the rise of analytical requirements, the evolution of distributed architecture and the popularity of open source model. We can see today that the database has changed from the initial hierarchical/network database to the relational database with a centralized architecture mainly oriented to OLTP requirements; to the OLAP data warehouse that is independent of the MPP architecture; and then to the more suitable for unstructured NoSQL databases and big data processing frameworks for mass data, distributed racks, and NewSQL, which combines the advantages of SQL and NoSQL at the same time. Facing the future, with the gradual maturity of new technologies such as cloud computing, 5G, and the Internet of Things, Huang Kaiyao believes that multi-mode and multi-state, computing power integration, and intelligent operation and maintenance will become new challenges for future databases.
According to Huang Kaiyao, in order to meet these challenges, openGauss 5.0 has undergone a major upgrade, and achieved a breakthrough in database technology through the innovation of the dual-engine kernel and architecture.
First of all, in terms of kernel innovation, the latest openGauss 5.0 version is compatible with the previous version. In addition to optimizing the performance of the entire link, technological innovations such as high intelligence and high security have been carried out to bring users better value of :
● In terms of high security, use the abnormal behavior detection technology based on the attention mechanism to accurately capture attack behavior; use the label-based access control to ensure the security of confidential data management and control, and improve the protection level.
● In terms of high performance, openGauss 5.0 has optimized the performance of the entire link, including the underlying operators, parallelism, concurrency, and the entire pooling layer, fully mobilizing resources and making full use of resources to achieve optimal performance. After optimization, multiple indicators increased by 30% to 40%. On the other hand, openGauss 5.0 implements global sharing, which greatly increases the amount of concurrency and allows the system to run quickly and stably.
● In terms of high availability, openGauss 5.0 has brought two important architectural breakthroughs this year: through shared storage, storage copies can be reduced to 1/3 of the original, greatly improving storage space utilization; through shared memory, read performance is improved, Achieve zero-latency reads. In addition, this version also brings a flashback function, allowing data to be rolled back to the previous moment to protect data security.
● In terms of high intelligence, database management is realized through the DBMind autonomous platform, and the third-generation intelligent optimizer ABO is upgraded to solve the pain points of traditional CBO/RBO in cardinality estimation and plan selection, making complex query plans better and performance improved by 30% %above.
Second, in terms of architectural innovation. openGauss 5.0 insists on being driven by user scenarios, and continues to innovate the architecture. Focusing on user needs and pain points such as multi-mode and multi-state, intelligent operation and maintenance, it launched the DataPod + DataKit combination to create a new data base.
First of all, DataPod can be divided into three layers: computing pooling, memory pooling, and storage pooling. Through software and hardware collaboration and full-stack optimization, a resource pooling architecture with full-stack observability, traceability, and out-of-the-box performance is finally realized. In terms of computing pooling, DataPod can optimize diversified computing power, including traditional X86 and Kunpeng computing power, and use the most suitable database engine to meet the data processing demands of different applications; in terms of storage pooling, it not only supports multiple In terms of memory pooling, it realizes memory interconnection between computing nodes and supports RDMA high-speed access, creating a transparent cross-node memory pool .
DataKit is positioned as the data lifecycle management platform of openGauss. It takes the plug-in architecture as the main body, defines the standardized interaction interface between plug-ins, and through community collaborative development, it covers six scenarios including deployment, development, and operation and maintenance.
值得注意的是，DataKit也是一个工具开发平台，还支持用户根据插件接口和规范实现自己的特有功能。功能插件支持动态加载到DataKit上运行，也支持从DataKit上动态卸载。插件支持调用平台和其他插件的开放接口获取相应数据和功能，以实现快速构建特色功能。
“未来，openGauss希望通过内核和架构的双引擎驱动，实现数据库技术突破，为大家带去一个更简单、更优雅、更开放、更智能的数据库的使用体验。”黄凯耀如是说。
openGauss开源三年来
如期兑现四个LTS版本
自2020年6月30日开源以来，openGauss数据库吸引了众多内核开发者的关注。如今已经开源了将近三年的时间，openGauss一直思考如何去构筑更好的国产数据库，如何在内核的架构和基础创新上面引领行业的发展。
在2020年发布的第一个openGauss基础版本中，就格外重视两大关键特性：一是高性能，在两路鲲鹏下TPCC Benchmark达成性能150万tpmc；另一个是高可用性，支持主备同步，备机并行恢复，10秒内即可升主提供服务。
在2021年发布的第二个版本中，openGauss在保持原有特性情况下新增了众多特性，特别是在高性能、高安全和智能化的打造上有了更大的突破，并提出了“四高”的内核，引领了数据库行业的发展。
到了2022年，openGauss 3.0针对应用场景进行重大升级，发布四项新的能力：包括面向边缘场景的轻量化版本；完备的集群管理组件；支持 MySQL 语法兼容和数据迁移；面向分布式应用场景，联合分布式中间件推出高性能的分布式解决方案。
最新发布的openGauss 5.0版本，不仅在内核上持续突破，还创新性地提出了DataPod + DataKit的解决方案：在计算池化、内存池化、存储池化三层上进行软硬协同，全栈优化；以插件化架构为主体，定义插件间的标准化交互接口，通过社区协同开发，覆盖部署、开发、运维等6大场景，打造数据全生命周期管理平台。
为了更好地满足行业细分场景的需求，充分发挥openGauss数据库和鲲鹏硬件软硬协同的优势，openGauss社区联合伙伴推出了“全栈可信”、“全栈安全”、“开箱即用”、“极简运维”的数据库一体机解决方案。这个数据库一体机基于openGauss 5.0的DataPod能力与开放的鲲鹏服务器硬件，性能做了专项优化，为客户提供开放、安全、自主创新、长期演进的企业级数据库解决方案新选择。
openGauss打造的DataPod+DataKit全新数据底座与持续优化的四高内核，不仅仅是在数据库技术发展延长线上的添砖加瓦，更是成为国内开源数据库技术发展的重要推动力量。
积极开源，良性生态助推
openGauss “枝繁叶茂”
数据库的发展离不开开源社区的支持。开源数据库的发展孕育了大量商业机遇，基于开源项目的创新商业模式层出不穷；同时商业活动的活跃可以反过来刺激开源社区壮大。事实上，开源开放、协同创新也一直是openGauss日渐繁荣的秘诀。
截至发稿前，openGauss开源社区已有290多家企业和机构加入，近5000名开发者参与社区贡献，下载量突破190万次。开源3年以来，openGauss已上线7个版本。DBV伙伴基于社区版发布商业发行版，用户基于社区版深度定制的用户自用版，这些版本已覆盖数字基础设施丰富的场景，包括面向关系型的集中式、KV数据库，地理空间数据库，时序数据库等。
为了提升openGauss社区专业服务能力，openGauss伙伴专业保障服务于openGauss 开发者峰会2023上正式发布。会上，共有八家伙伴成为openGauss社区首批认证服务伙伴，他们是openGauss生态服务能力构建的有力保障，承诺为用户提供高质量、高可靠和稳定的专业服务。
在笔者看来，作为数字平台的核心软件之一，数据库具有非常重要的战略意义，需要长期的投入、技术积累，才能有经受得住考验的产品。openGauss社区通过持续发展、不断繁荣的社区生态，完善了我国在数字建设过程中的关键一环，加快了我国自主创新建设。未来，将持续为客户带来价值，为国内各个行业提供数据库的支撑，为我们国家数字化转型提供强有力的底座支持。
openGauss开源三年来的快速发展，离不开每一位全产业链伙伴的支持和贡献。面向未来，虽道阻且长，但行则将至，也希望openGauss社区可以充分联合社区创新力量，共研核心技术、共创应用示范、共推行业标准、共创国产数据库发展新生态。
本文分享自微信公众号 - openGauss（openGauss）。
如有侵权，请联系 [email protected] 删除。
本文参与“OSC源创计划”，欢迎正在阅读的你也加入，一起分享。