From Ali cloud database into Gartner on the evolution of database

According to the latest research report of Gartner, a global authoritative IT consulting company, in the magic quadrant of database system in 2018, Ali Cloud database was included in the “visionary” quadrant, which is the first time that domestic database entered Gartner magic quadrant. Gartner’s magic quadrants describe the product capabilities and market size of database vendors. The four quadrants are leaders, challengers, visionaries and domain specific. At present, Oracle, Microsoft and IBM are the leaders of database magic Quadrant. Ali Cloud through continuous technology research and development innovation, into the visionary quadrant, on behalf of the company has good product technology innovation and forward-looking, after further breakthrough in the market will hopefully become the absolute leader in this field.

“Ali cloud have all cloud vendors in the database of the most abundant species coverage, almost for all customer business scenario offers a variety of solutions” analysts wrote in a report, “in addition to the public cloud, ali cloud database also provides a complete privatization deployment form Apsara solution Stack and hybrid cloud data sharing, This is an important advantage that AWS and Google lack. According to Gartner’s market analysis report, Alicloud database ranks first among all cloud computing vendors in China. Gartner pointed out that Alibaba’s open source AliSQL has a solid user base, especially the improvement in management and security has been widely recognized. In addition, the application of cloud native POLARDB in high performance and high capacity scenarios is worth looking forward to.

As the only domestic database manufacturer, Ali Cloud was selected into the “magic quadrant of global database” for the first time, we further interpret this milestone event through in-depth analysis of the evolution of cloud database.

Trends: Evolution of cloud databases

As one of the core technologies of IT science and technology, database originated in the 1970s. IT has survived for a long time in the tide of technological development and supports the economic activities and commercial civilization of modern society. From Kevin Kelly’s “Out of Control”, “What Technology Wants” and “Inevitable”, to Yuval Harari’s recent trilogy “A Brief History of Mankind”, “A Brief History of the Future” and “A Brief History of Today” are both widely followed by the public. These works of science and technology fully reflect today’s society for the future trend of science and technology and human society’s full prospects and reverie. Both social individuals and economic enterprises try to gain insights through perception and imagination of future trends, and the key word is “evolution”. Similarly, the difference between cloud database and traditional database can be fully reflected in the evolution of traditional database to cloud technology integration.

The rapid development of cloud computing is inseparable from the Internet. The first opportunity for cloud computing services to incubate is the Internet e-commerce giants to make full use of a large number of idle IT servers such a background. At the beginning of the Internet developers and entrepreneurial companies to provide easy-to-use, fast, cost-effective IT resource services. To put a database on the cloud, solve the resource sharing problem among multiple tenants, as shown in Figure 1-1. By dividing and isolating the CPU, memory, local storage, and network resources of physical servers through virtual machines or containers, multiple database instances can be run to serve multiple cloud users. Of course, high-speed SSD storage is also standard to reduce crosstalk of I/O performance between multiple databases. This is the earliest form of cloud database, characterized by the best cost, and is still used by some developers and small customers who are primarily concerned with cost.

Figure 1-1 Phase 1: Cloud database based on virtualization and local SSDS

With the business development of these small customers, some small startups gradually become unicorns. Some mission-critical Critical business applications need to provide 7X24 hours of uninterrupted service, so the new form is evolving. On the basis of the first stage, by adding a standby warehouse, as shown in Figure 1-2, Another piece of data is synchronized between the master and the standby, and then only the master provides services, and the standby takes over from the master whenever the master fails. With this active-standby architecture, 99.95% high availability services can be achieved. This is a simple and practical high availability mechanism proven by many production environments. But at the cost of adding a hidden standby machine, the data is also doubled.

Figure 1-2 Phase 2: Highly available cloud database

After the high availability problem is solved, the system load increases with the growth of user services and data. A practical read/write separation architecture is introduced, as shown in Figure 1-3. In most database application scenarios, the ratio of read and write requests is more than 3 to 1, and in many scenarios, the ratio of read and write requests is 10 to 1 or higher. In this case, multiple read-only nodes (read-only libraries) are introduced to synchronize one copy of the master node’s data on each read-only node to share the read load. The overall throughput capacity of the system is improved several times, which can effectively solve the majority of concurrent processing scenarios with large data volumes. Of course, due to the introduction of read-only nodes, the high availability of multiple read-only nodes also requires a hidden standby library to support, and an adaptive load balancing read-write separation proxy interface for data connection and automatic distribution. As a result, the overall processing power of the system increases, but the cost increases rapidly. For example, if there are N read-only nodes, the total system data has N+1+2 copies (1 primary node, 2 standby repositories). In addition, the maximum amount of data supported by a dedicated server is 2TB to 3TB due to local storage limitations. Therefore, the appeal architecture needs further evolution in terms of concurrency performance, data capacity, cost, elasticity and so on.

Figure 1-3 High availability cloud database with read/write separation architecture

Figure 1-4 shows a qualitative leap in the evolution of cloud databases. A high-availability read/write separation architecture based on shared storage, which utilizes distributed shared storage and three-copy data storage to improve data reliability. Using the data sharing architecture, read-only nodes do not need to replicate data, which greatly improves system performance. In addition, the system supports up to 15 read-only nodes and ensures a lower data delay for read-only nodes. When there is only one read-only node, the read-only node can obtain read/write separation capability and provide double read performance. Compared with the former active-standby high availability mechanism, the active-active high availability mechanism is an improvement in technological evolution. In short, based on shared distributed storage will bring essential technical architecture advantages, Ali cloud cloud native database POLARDB, is based on such a kind of latest technical architecture evolution and development, according to the 28 rules and combined with the actual operation and maintenance experience, can meet more than 80% of the user relational data processing scenarios.

Figure 1-4 Cloud native database with high availability read/write separation based on distributed shared storage

From the perspective of the evolution process of appellate database of cloud, the cloud database expand the service ability of the enterprise overall scheme (high reliability, high availability and load balancing, separation, speaking, reading and writing, etc.), and on the basis of the ability of a cloud platform, through the evolution of the technical architecture form a cloud native database, provides a superset of traditional database ability, The overall cloud service system from resources, platform to ecology, as shown in Figure 1-5, has the following characteristics compared with traditional databases:

1. Cloud database further expands the computing capabilities of traditional databases through virtualization and distributed technologies of computing, network, and storage. It also provides fast and flexible service capability that can be changed according to users’ business needs.

2. Cloud database provides time-saving, worry saving and labor-saving IDC hardware deployment, monitoring management and software upgrade and maintenance services based on green computing, security and high-speed network.

3. The cloud database greatly simplifies the purchase, operation, and management of users, and greatly reduces the time and application costs through web-based self-service management.

4. Cloud database directly provides other external software and hardware tools and modules for traditional on-premise critical applications that depend on high availability, load balancing, firewalls, network topology, cross-room DISASTER recovery, etc., to build secure and reliable database service capabilities that meet enterprise solutions.

Figure 1-5 Aliyun Apsara cloud database service system

Cloud native relational databases: a source of value for enterprise-level information

For modern enterprises, they need to quickly explore new values and opportunities in enterprise cloud computing data. IT resource capabilities provided by cloud services have become the best practice of the overall digitalization plan. The competition for productivity in the future is no longer just about horizontal production scale, operational efficiency, and access to resources. More importantly, we need to provide more lean productivity and accurate marketing ability vertically in response to rapid market changes in user needs. In the digital era, enterprises need to be able to adapt quickly and innovate. Through data association and analysis, enterprises can quickly, directly and accurately perceive demand and discern market changes, and create personalized, refined and novel products and services, bringing new vitality to enterprises. Digital IT architecture gradually become the standard enterprise cloud computing, cloud services has not meet the most early the form and the ability of small and medium-sized Internet users, cloud native relational database to around the enterprise cloud applications focused on business continuity, high performance processing capabilities, safety and reliability, extensibility and elastic ability to provide the overall solution and service ability. As shown in figure 1-6, enterprise-class cloud applications on multiple dimensions of data management, information and data management and business process, for data processing, the trading process, analysis, extraction, finally settle such as workflow is the most precious, can directly express user information, demand information, business performance and business value of multiple modules, Are closely associated with relational cloud databases.

Figure 1-6 Enterprise cloud data and information architecture

POLARDB: 100% MySQL compatible relational native database

Compared with the early cloud database services, cloud native database POLARDB can provide enterprises with better security and reliable characteristics, better business continuity, faster business demand strain ability, better scalability, of course, and higher cost performance. POLARDB uses the latest software and hardware integration of innovative fusion design, in a number of technical indicators to achieve the world’s leading.

1. Extreme speed performance

POLARDB’s computing and storage engines are optimized to handle enterprise-class large-scale data processing with read performance of up to 1 million QPS and write performance of more than 250,000 QPS through standard Sysbench OLTP performance benchmarks. The system supports a maximum of 16 processing nodes and achieves a maximum concurrent aggregate read performance of 10 million QPS. In addition, POLARDB provides acceleration module for data complex query ability, which can obtain up to 4-20 times of query acceleration ability.

2. Super capacity and elastic expansion

POLARDB adopts a distributed shared storage architecture, separating computing and storage, using the most advanced 3D Xpoints storage media. The storage space is automatically expanded according to users’ actual use data, and the storage is charged according to users’ actual usage. No matter how much data, within 5 minutes can be completed node specifications upgrade and read-only node expansion, as needed, on demand, fast adaptation.

3. 100% compatible with MySQL

100% compatibility with MySQL OLTP (online transaction processing) and downward compatibility with MySQL5.6 greatly reduces the cost of user service adaptation and minimizes the cost of cloud adaptation and transformation risk on the enterprise.

4. High reliability and availability

POLARDB uses a distributed mass storage system with high data reliability through multiple replicas. The underlying storage data can be backed up by Snapshot in seconds. In addition, it supports strong and consistent same-city DISASTER recovery capability, and provides the most reliable technical support for the core value data of the enterprise.

5. Data security

POLARDB uses white list, VPC network, SSL encryption, SQL audit, data multiple copy storage, backup and recovery and other comprehensive means to provide security for database data access, storage, management and other links. Enterprise-level data can be secured in the cloud.

We believe that in the future industry integration, cross-border attempts, minimally invasive new applications, enterprises will continue to insight and explore the value of data. Ali Cloud database, through continuous innovation of technical advantages, will provide the most direct help for the rapid evolution of lean productivity.