2022 is bound to be full with uncertainties, with recurring epidemics, a sluggish economy, effected global supply chains, energy and food crises and so on. All challenge the survival and development of enterprises. The front pages and headlines of Internet news are saturated with shrunk market, broken capital chain and corporate layoffs. Facing so severe a situation, how can enterprises step out of the mire?

　　The key to it is digital transformation still, among which big data industry is the first priority.

　　As one of the most critical infrastructures of the big data industry, the importance of the database is self-explantory. Take three major operators, China Mobile, China Unicom, China Telecom, as an example, thousands of industries, tens of millions of enterprises and hundreds of millions of people all over the country are enjoying the connectivity services and convenience provided by the operators. In order to ensure the secure network environment and business operation needs of major telecom operators, AsiaInfo’s AntDB plays an extremely important role in it.

　　AntDB was started in 2008. On the core system of operators, it provides online services for more than 1 billion users in 24 provinces across the country. AntDB can process one million core communications transactions per second at its peak, ensuring the continuous and stable operation of the system for nearly ten years with 0 failure.

　　Because of this, how to comprehensively improve the performance and efficiency of the database system and reduce the cost of expenditure has always been one of favorite research topics of the major operators. Especially in the face of the impact of the epidemic, how to use the database for enterprises to reduce costs and increase efficiency, has also become the focus of the majority of enterprises recently.

Strong alliance of AntDB and AMD EPYC

　　Recently, in the field of domestic database, there is something that attracts the attention of many people in the industry. That is, in order to bring an overall improvement in performance and computational efficiency to AntDB system, and at the same time significantly reduce cost and power consumption, AsiaInfo has chosen to cooperate deeply with AMD and equip AntDB with AMD's third-generation EPYC processor with the latest architecture.

　　Why did AsiaInfo choose AMD EPYC? I'm so excited to tell you about it!

　　As far as I’m concerned, the AntDB system uses AMD EPYC 7003 series processors, which are based on 7nm advanced process technology and AMD Infinity architecture, with up to 64 high-performance "Zen3" x86 cores, 128 threads, and 256MB L3 Cache. With AMD 3D V-Cache? technology, this series of CPUs has a 768MB L3 cache! To date, AMD EPYC has not only set a number of world firsts for x86 data center CPUs, but has also set more than 250 performance world records:

　　1. 7nm advanced process technology.

　　2. up to 64 cores and 128 threads.

　　3. 4.10 GHz of maximum accelerated clock frequency.

　　4. 8 high-speed memory channels, supporting DDR4-3200.

　　5. Support for up to 4TB of memory capacity and 8TB for dual circuit systems.

　　6. 128 PCIe 4.0 high-speed I/O lanes.

　　7. Up to 768MB of Level 3 cache (AMD 3D V-Cache?).

　　8. Hardware-level SME/SEV and other security encryption technologies.

　　Microarchitecture of AMD EPYC 7003 Series Processor

　　In addition, AMD EPYC 7003 series processors have a 19% increase in the third generation "Zen3" core IPC (Instructions Per Cycle) over the previous generation; an optimized L3 cache architecture that enables 32MB of L3 cache to be shared by every 8 cores, doubling the cache capacity per core; support for 4-6-8 memory interleaving for better cost performance in multi-DIMM configurations; and the security processor on CPU's integrated chip that supports industry-leading security features such as Secure Memory Encryption (SME), Secure Encrypted Virtualization (SEV), and Secure Nested Paging (SEV-SNP), with hardware-level security encryption technology to protect critical customer data.

　　In short, as the latest generation of AMD's server CPUs, AMD EPYC 7003 series processors' powerful computing power, massive I/O expansion features, and hardware-level security encryption technology can provide strong support for AsiaInfo's AntDB system, which is a perfect match between the two sides.

Logical Architecture of AMD EPYC 7713 (64 Core, 2.0GHz) CPU

AntDB with long history and many users

　　As an independent, innovative, secure and controllable database product created by AsiaInfo , AntDB is a general-purpose enterprise-level native distributed relational database with high availability, high performance and native distributed architecture, which realizes online elastic scaling and distributed strong consistency and can exponentially improve the data processing capacity of the system, to help customers easily achieve petabyte-level data management, and promote the development of the relational database from traditional centralized storage and computing to the aim of being distributed, intelligent and full-stack.

　　As the first batch of domestic database, AntDB was put into use in mobile billing system as early as 2008, and now it is widely used in communication, finance, transportation, energy, Internet of Things and other industries, and has been successfully implemented in more than 200 projects of China Telecom, China Mobile and China Unicom.

　　In addition, AntDB is also a home-grown database serving a large number of users. Up to now, the number of users served by AntDB directly or indirectly has exceeded 1 billion. Whether people call or surf the Internet, they can't live without AntDB's service.

　　As a member of first-tier domestic database with its product features such as high performance, elastic expansion and high reliability, AntDB can process one million core communications transactions per second at its peak. As we all know, among all industries, the carrier business has the highest demand for database performance. In the past ten years, AntDB has guaranteed the continuous 0-failure smooth operation of carrier core database, which fully illustrates that its reliability and stability are top in the whole industry.

The architecture of AntDB

　　Since the AntDB system is so important, let's imagine how much operators and users would gain if they could improve their computing power and efficiency, or reduce their cost and power consumption?

　　The key lies in the third generation of AMD EPYC series processors.

The innovative features of AntDB combined with AMD EPYC

　　Thanks to the new processor architecture of the third generation AMD EPYC, the strong partnership between AsiaInfo and AMD also brings a series of innovative features to the AntDB:

　　1. On the dual EPYC system, up to 128 cores and 256 threads of super computing power can fully utilize the high concurrency capability of the CPU, so that the application side can obtain exponential performance improvement.

　　2. AMD EPYC SoC architecture, which naturally fits the containerized deployment of AntDB system, significantly improves the processing efficiency of core transactions such as billing statements and business analysis, and improves customer experience.

　　3. As a carrier-level core transaction database, AntDB not only requires more cores of concurrent computing power, but also is very sensitive to the latency and bandwidth of memory access. AMD Infinity architecture two-way EPYC system has 16 high-speed memory channels and bandwidth, and supports up to 8TB of memory capacity.

　　For the AntDB equipped with the third generation AMD EYPC series processors, AsiaInfo and AMD have conducted comprehensive optimization and performance evaluation tests in both hardware and software. AntDB was deployed based on single EPYC single AntDB instance and dual EPYC multiple AntDB instance, and the performance of the whole system was tested by using BenchmarkSQL and DBGen tools such as TPC-C and TPC-H industry standard models.

AntDB system performance

　　The test results show that, thanks to the multi-core concurrency capability and architectural advantages of AMD EPYC series processors, the single-way EPYC server running AntDB single database instance achieves 1.65 times the performance of the original dual-way 20-core configuration single instance, which really helps customers achieve a leap while further optimizing hardware costs; while the performance of dual-way EPYC multi-database instance deployment is doubled on the basis of single-way. In distributed mode deployment scale, AntDB achieves good multi-node linear scalability based on AMD EPYC processor platform.

Distributed AntDB Multi-Node Scaling Performance

Benefits for enterprise customers

　　The previous section introduced the many improvements and new features brought by AntDB combined with AMD EPYC, but for enterprise users, the most important question is what benefits can the two sides bring together? We can summarize them into three parts:

1. Higher computing power and computational efficiency

　　Thanks to the unique memory-disk dual storage engine and other features of AsiaInfo’s AntDB, as well as the third-generation AMD EPYC processor's larger L3 cache and IPC enhancement, higher memory bandwidth, higher branch prediction accuracy, up to 64 cores of concurrent computing power, and a high-performance database solution combining hardware and software, it can provide more computing power and computational efficiency for various business scenarios of enterprise relational database.

2. Lower TCO total cost of ownership

　　Thanks to the super performance of the third-generation AMD EPYC processor, AntDB can achieve the performance of the original dual-way server based on single-way EPYC, and further double the performance of dual-way EPYC multi-AntDB database instance deployment, realize the intensive deployment of higher computing power density, significantly reduce the hardware cost and the comprehensive cost of cabinet space, power consumption, electricity, etc., and help enterprises achieve cost reduction and improved efficiency.

3. To help achieve the "double carbon" goal

　　Thanks to the industry-leading process technology and processor microarchitecture, AMD EPYC achieves a breakthrough in both performance and computing efficiency, not only significantly reducing average power consumption per performance, but also significantly reducing server room energy consumption through intensive deployment, helping enterprises build green data centers and accelerating the achievement of "dual carbon" goals(carbon peaking and carbon neutrality goals).

　　No matter how IT technology develops and evolves, more computing power, less cost and lower energy consumption are always the unremitting pursuit of enterprises. Based on the third generation AMD EPYC processor,  AsiaInfo’s AntDB has achieved surprising results in improving system performance, reducing overall cost and reducing carbon emission. If you also want to make your database like a tiger, then I believe the success story of AntDB has provided enough inspiration.