Carrier-class servers are crucial for digital transformation

carrier-class servers
Image by monsitj | Bigstockphoto
Carrier-class servers ZTE digital transformation
Wang Quan

In this interview, Wang Quan, Vice President of Telecom Cloud and Core Network Products at ZTE, explains why carrier-class servers are essential to support digital transformation and why they need to be customizable.

Disruptive.Asia: What are the big trends driving innovation in the server market right now?

Wang Quan: Certainly, the rapid rise of emerging areas like artificial intelligence, IoT and metaverse are raising the demand for computing power. This is especially true of AI and machine learning, which are taking data analytics to new performance levels. Related to that, we’re seeing explosive data growth, the ubiquitous distribution of data resources, and a continuous decrease in the cost of data applications per bit.

Meanwhile, the ongoing digital transformation of industries is also driving the growth of the server market. Servers are the core components and supporting engines of digital transformation because their quality determines the continuity and stability of enterprise services.

All of this is creating new scenarios and requirements that are driving the server industry, and these are constantly emerging and evolving. This also creates new challenges for the server and storage market in terms of performance, reliability, and innovation.

Recently ZTE released its new G5 server series in Thailand– what do they bring to the table?

ZTE has released five types of new servers, including the 5200 G5 high-density server, R5300 G5 full-scenario server, R5500 G5 mass storage server, R6500 G5 heterogeneous computing power server, and R8500 G5 high-performance server.

Some of the key features include liquid cooling for heat dissipation, high-density computing power, mass storage, and stability and reliability.

Liquid cooling is a key feature because the heat dissipation requirements of chips are rapidly increasing, especially as high-computing-power GPUs become more widely used in IDCs. The limit value of the air cooling technology is generally between 300-350w, which is far from enough to bear the energy consumption of computing power in the future. Liquid cooling will be a mandatory solution for high computing power IDCs to dissipate heat.

For the R6500, the server has 10-20 built-in intelligent acceleration engines for heterogeneous computing, which can flexibly schedule heterogeneous computing power and provide flexible expansion capabilities. It can also be customized in accordance with application scenarios. I should add that all of the new G5 servers support flexible expansion.

The server market is an increasingly crowded one, partly because of the trends you’ve just mentioned – how does ZTE differentiate itself to stand out?

For one thing, we have a powerful self-development capability that enables us to launch new server products based on the latest computing power platform as soon as possible. Key components such as the mainboard, BIOS, BMC and NIC are independently developed, and we can support computing power platforms such as Intel, AMD and Hygou.

Also, ZTE ensures carrier-class reliability of servers through strict standards and specifications that run through the entire process of component sourcing to product delivery. At the R&D stage, the components comply with the rigorous standards for telecommunications component sourcing to deal with extreme environments effectively. Each product must pass more than 10000 verifications and tests higher than industry standards, such as the thermal test at –60°C to 80°C and the drop test with a height 10 cm above the industry standard, eliminating potential risks during the design phase.

On the manufacturing side, we launched an intelligent production line in 2022 with an annual output of 500,000 servers. Thanks to 5G, big data, and AI, innovative technologies such as machine vision, robotic arms, and AGVs become a reality in the production line, ensuring a safer workplace and higher productivity.

During the pre-delivery inspection, each server must undergo the 24-hour high-temperature aging test before delivery.

According to IDC, ZTE servers business ranks No.1 consistently in the Chinese telecom industry with a market share of over 30% and ranked fifth in China and ranked eighth in the global market.

Technical prowess is one thing – how is your customer service support?

ZTE has customers in over 160 countries and regions, and we provide 24/7 technical support services to all of them to ensure the secure and stable operation of networks and services. We have been doing this for decades. Based on that experience, we have established a management system for project delivery and after-sales services involving an efficient project management and operation mechanism, a real-time intelligent digital system, and a high-performing team.

In terms of the supply chain, ZTE works with partners to ensure balanced, efficient, stable, and sustainable supply of servers.

ZTE also has an advanced after-sales service system, with over 10,000 engineers in 107 countries providing “factory services” for all customers worldwide. Additionally, we are the only manufacturer capable of end-to-end integration of servers, storage products, network devices, cloud platforms, and upper-layer applications. That means ZTE is better positioned to provide tailored solutions and products for customers to meet the requirements of different industries.

What are the next big areas of innovation that you’re pursuing in the server segment?

There are three key areas we’re focused on. The first, of course, is heterogeneous computing, which is featured in the new G series. Heterogeneous computing is inevitable because complicated application requirements cannot be met by solely relying on general-purpose X86 CPUs. We need acceleration hardware components, such as smart NICs, GPUs, FPGAs, and even neural processing units (NPUs).

ZTE is working on the new data processing unit (DPU)-centric computing architecture, where DPU takes over the I/O load from the CPU. In addition, like a switch, the DPU can connect to multiple XPUs through a software-defined data bus to better meet the requirements of services featuring high throughput and dense computing, such as video processing, AI-driven recognition, and autonomous driving.

The second is distributed architecture, which is crucial as more businesses become data-driven. ZTE has developed a distributed disk array product by giving full play to the advantages of horizontal scaling of distributed storage as well as low latency and high performance of traditional disk arrays.

And the third area of innovation is comprehensive energy saving. Obviously, the power requirements of servers is growing, so the server architecture needs to be designed to process the largest amount of data with the lowest power consumption. We’ve optimized the power supply for servers, which reduces total server energy consumption through intelligent CPU frequency adjustment, automatic HDD hibernation, and intelligent power and battery management.

Furthermore, ZTE is exploring new architectures, such as in-memory computing, and new materials, such as RRAM and PCM, which could, we estimate, reduce power consumption by more than 50%.

Related article: ZTE unveils more efficient, eco-friendly products at MWC

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.