Author: Community Assistant Original source: https://tidb.net/blog/1ff45c7d
Thirty years ago, to write code, programmers had to use complex assembly language.
But today, few programmers know how to use assembly language, and simpler high-level languages such as C++, C#, JAVA, Rust, and Go have become the mainstream of development.
As the technology wave changes, each generation of developers uses technology in a completely different way than the past.
Just like the change of development language, database technology, which is the core cornerstone of IT system, is also evolving from complex to simple.
At the recent PingCAP DevCon 2022 conference, Huang Dongxu, co-founder and CTO of PingCAP, said that future developers may not know what database technology is, and they don’t need to care about what database technology is, because Serverless is reducing the complexity of database technology. Make technology insensitive.
In fact, the concept of serverless has become very popular in recent years. Well-known consulting organizations such as Gartner and Forrester have paid attention to serverless. Big cloud computing companies such as AWS, Alibaba Cloud, and Tencent Cloud are also continuously deploying serverless related products. It can be said that the combination with Serverless has once again fueled the development of the database.
So, what exactly is a serverless database and what is its value? Going a step further, will Serverless become the future form of databases?
Database to Serverless
If you want to understand the serverless database, you must first understand the development process of the database.
In the early days, users generally built their own databases. As the most traditional database application method, users need to consider all aspects of physical deployment and operation and maintenance when deploying in their own computer rooms. Traditional databases also have low flexibility and scalability, and are expensive and expensive to maintain.
With the popularization of cloud computing, the application of databases to the cloud has become the mainstream. The database is provided on the cloud in the form of PaaS service and leased service, and users no longer need to care about the physical deployment of the computer room.
Because the cloud database at this stage simply migrates the database from the local to the cloud without making too many changes in the architecture, the flexibility of the database is still limited.
To solve this problem, cloud-native databases emerged.
Because the cloud-native database is completely designed for the cloud, it completely decouples computing and storage resources, uses distributed cloud storage instead of local storage, and makes the computing layer stateless, so that it can give full play to the advantages of the cloud and has the characteristics of elasticity and scalability. So that users do not need to worry about daily business expansion.
However, cloud-native databases also have their own bottlenecks, that is, they cannot be automatically scaled on demand, nor can they be paid according to usage at a smaller granularity. When users encounter a sudden demand for database expansion, they can only manually adjust the database capacity according to the actual business usage.
Although this approach is indeed feasible, it will consume a lot of time and cost. Even for database experts, it is not an easy task to manually manage database capacity in the face of fluctuating applications, taking performance and cost into consideration.
In this context, Serverless databases emerged.
Due to its fully automatic expansion capability, the serverless database can intelligently “expand” and “shrink” as the number of user business requests increases and decreases, and realize automatic “throughput” of resources. When the traffic peak comes, resource support can be automatically allocated; when the traffic enters a low valley, resources can be automatically released to save costs.
This ability is the embodiment of the concept of serverless, which abstracts the bottom layer of the database and the irrelevant parts of the business, and provides developers with a direct operating environment, so that developers can directly call the function platform to complete the function operation without caring about the server infrastructure. .
The logic and state of the server are also managed by the service provider, and the service will only be automatically scaled when needed, so that the database can achieve extreme elasticity, and developers no longer need to be troubled by complex underlying infrastructure.
The Value of Serverless Databases
The popularity of the concept of serverless has brought more and more attention to serverless databases.
In fact, from the perspective of developers, it is not difficult to understand why serverless databases have been widely sought after as soon as they appeared.
A survey titled “Who is choosing Database in your organization” shows that architects, developers, and DBAs, as real users of database software, spend 41% of their daily working hours doing basic work. Facility maintenance, such as buying servers, deploying servers, operation and maintenance, etc., only 39% of the time is spent on business innovation.
As the data structure becomes more and more complex and there are more and more databases, each database has its own set of technologies, and developers have to learn more and more things. To solve this complexity, release the productivity of developers, and allow them to focus more on business innovation, the abstraction brought by Serverless is essential.
According to Huang Dongxu, co-founder and CTO of PingCAP, the higher the level of abstraction, the higher the development efficiency.
From traditional databases to the cloud to cloud-native databases, cloud infrastructure capabilities and database kernel-level capabilities have been abstracted step by step, allowing the database to support efficient application development iterations. Today, Serverless is further abstracted on the basis of cloud native, which can further improve development efficiency.
Huang Dongxu, co-founder and CTO of PingCAP
Therefore, Huang Dongxu believes that Serverless is a development model evolved after cloud native matures, and serverless database is the inevitable result of the development of cloud native database .
“For all innovative database companies, if the ticket for the first two years was cloud-native, then this year’s ticket will become Serverless,” Huang Dongxu said.
With a high degree of abstraction, the value brought by the serverless database is obvious:
**First, it is easy to create. **For the creation of a Serverless database, users don’t need to care about any deployment details. It can be created with one click within tens of seconds.
**Second, automatic scaling. **Users do not need to consider the infrastructure, and the serverless database can be automatically matched according to changes in business load. When the business throughput reaches a certain level, there is no need to stop to add servers, and the system will automatically expand; when the business peak drops, the system can automatically shrink, or even shrink to 0.
**Third, cost saving. **Serverless databases can provide finer-grained billing, paying according to actual usage, and not billing if you don’t use it.
**Fourth, deep integration with application development experience. **In the past, databases only cared about various indicators such as performance and stability, and were rarely designed from the perspective of developers.
The emergence of Serverless allows databases to truly integrate into the modern development and application process from the perspective of users, helping users build applications faster and more smoothly.
Take the TiDB Cloud Serverless Tier released by PingCAP as an example. As an online, fully automatic HTAP Serverless database service, developers can deploy and configure a fully functional Serverless TiDB database with just a few clicks, which can satisfy users Automatically scale up and down on demand without having to manage server infrastructure.
At the same time, developers do not need to prepay for TiDB Cloud Serverless Tier, and can pay according to the actual SQL requests submitted and the actual storage used, which is a cost-effective database usage model.
In practical applications, TiDB Cloud Serverless Tier has also proved its worth. It used to take 15 minutes to start a new TiDB cluster on the cloud, and it may take longer if the user deploys it himself, but now it only takes 20 seconds to start TiDB Cloud Serverless Tier. At the same time, TiDB Cloud Serverless Tier also reduces the cost of TiDB on the cloud, and only the smallest cluster reduces the cost to 1/5.
“Imagine, if the startup time is shortened to three to five seconds, and you don’t need to care about the scalability of it (TiDB Cloud Serverless Tier), even if the business traffic becomes huge, it can scale up very well. When there is no traffic , it can also retract automatically, which will greatly change the use process and experience of developing applications,” Huang Dongxu said.
Serverless databases go to thousands of industries
As a cutting-edge technology, Serverless is not only exclusive to technology companies or large enterprises, but also the benefit of small and medium-sized enterprises. Serverless databases can be widely used in business scenarios in various industries such as e-commerce, FMCG, finance, and telecommunications.
According to Huang Dongxu, co-founder and CTO of PingCAP, since the launch of TiDB Cloud Serverless Tier one month ago, most of the users except for large enterprises are small and medium-sized waist users, and they can quickly activate and use the service themselves.
Interestingly, this group of users is not the core users of previous databases, such as architects, DBAs, etc. They don’t know much about database technology, but they can quickly use Serverless databases.
This happens to be the best value embodiment of serverless technology – easy to use and cheap, so that the database will always be the pursuit of users.
As a database innovation manufacturer with more than 3,000 enterprise users around the world and operating its own open source community, PingCAP always puts “meeting user needs” in the first place, and now the rapid launch of Serverless database is also a keen insight into user needs and respond.
In fact, serverless technology is still in its early stages, and there are not many vendors who can build serverless databases. Basically, they are concentrated in public cloud vendors, who use serverless databases to match their own cloud basic service business. As an independent third-party database vendor, PingCAP was able to take the lead in launching a serverless database for two reasons:
First, leading technology and continuous innovation.
Continuous innovation is the principle that PingCAP has been adhering to for a long time, and it is also one of the secrets for PingCAP to maintain its competitiveness.
PingCAP founder and CEO Liu Qi once talked about PingCAP’s understanding of innovation: “TiDB’s evolution process in the entire architecture is also gradually staged. The earliest TiDB is called NewSQL. Following the pace of PingCAP, a large number of databases have emerged in the industry called NewSQL. NewSQL; PingCAP proposed the HTAP architecture in 2017, and later databases also mentioned HTAP.”
With its leading HTAP technical capabilities, PingCAP was selected into the “Forrester Wave: Translytical Data Platform, Q4 2022” report and entered the “Strong Performers” quadrant, becoming the only Chinese database vendor shortlisted.
As a cloud-native distributed database based on the HTAP architecture, TiDB itself is a separate storage and computing architecture. With the advantages of the cloud, it can achieve flexible resource scaling, elastic scaling, and business continuity guarantees, and supports users on the cloud. The database runs key business transactions and real-time analysis tasks at the same time, and this leading technical architecture also lays a solid foundation for the evolution to the serverless database.
In terms of the original cloud-native features, TiDB makes full use of the infrastructure on the cloud and upgrades and transforms the cloud-native at the architecture level.
On the one hand, the shared storage of cloud vendors is introduced to realize the upgrade from the Shared nothing architecture to the Shared everything architecture at the TiKV layer.
On the other hand, in order to reduce the complexity of the database system, the micro-service transformation of the database allows any separable components to be serviced (API calls), thereby creating a serverless database.
It is not difficult to find that PingCAP is pursuing a kind of “simplicity” engraved in the bones. When solving the difficult problems faced by all users in the database, it will leave the complexity to itself and give simplicity to customers.
Second, around open source, embrace users.
From the beginning of its establishment, PingCAP has taken open source as its core strategy, and open source is the gene of PingCAP.
Based on the strong “vitality” of the open source community, PingCAP has 1895 developers distributed in 45 countries and regions around the world. Among them, contributors other than PingCAP accounted for more than 75%, of which nearly 20% were corporate decision makers.
Because of being closely related to the open source community, PingCAP can listen to developers very well, obtain the ultimate user scenarios in a timely manner, think what users think, and do what users need.
Through global community technical collaboration, PingCAP can quickly iterate TiDB products and ecology, and truly deliver product value to customers.
The ultimate direction of the database is technology-insensitive
As one of the important trends of cloud computing in the next ten years, Serverless has shown great potential, and Serverless database is becoming the general trend.
As Wei Kai, deputy director of the Cloud Institute of the China Academy of Information and Communications Technology, said, in today’s global economic downturn, enterprises need to manage growing data assets at a lower cost and with higher efficiency. How to make better use of cloud infrastructure For facility services and more refined management of databases on the cloud, Serverless is the direction of the future.
But is Serverless the final form of the database? Where will the serverless database go next?
According to Huang Dongxu, co-founder and CTO of PingCAP, the future direction of the serverless database is to provide a smooth development experience and improve the happiness index of every application developer. “The database should be a very simple thing. Developers don’t need to understand how the bottom layer of the database is implemented and how to use it well. Instead, they can spend more time focusing on business innovation and those really important things.” .
Going a step further, Huang Dongxu believes that technology insensitivity will be the ultimate goal of the database and the entire infrastructure software. “In the future, developers should only care about the API for data manipulation, which is closer to the business. On the other hand, when Serverless is provided on the cloud, data sharing and exchange become a It is very natural or very simple, at that time there may be a new business model called ‘Data market’.”
Conclusion
With the spiraling rise of technological evolution and industrial development, Serverless is on the stage of history, leading the development of cloud in the next decade. From a global perspective, if today’s databases have not taken the serverless route, they will completely lose their leadership in the next few years.
As the future form of database, serverless will eventually bring ultimate simplicity and value. This process requires not only the establishment of users’ minds, but also the continuous improvement of product capabilities by leading service providers such as PingCAP.
As Huang Dongxu said, “Technology ineffectiveness is our ultimate direction of progress, and it is also our responsibility as a basic software provider.”
The knowledge points of the article match the official knowledge files, and you can further learn relevant knowledge Cloud native entry skill treeHomepageOverview 11124 people are learning systematically