In the 5G era, with data volumes surging, what are the related data storage and processing implications?
In 2019, global 5G network construction was in full swing. Dozens of top global telecoms operators have announced the commercial use of 5G. More services such as autonomous driving, smart manufacturing, smart healthcare and immersive reality will be introduced. Global data volume is expected to grow rapidly from 33ZB in 2018 to 180ZB in 2025. Traditional IT infrastructure will face major challenges, including the need for massive data storage and low-latency, real-time processing.
We can classify all the data that operators have into three categories: production and transaction data; massive video data; and edge data. Each type has its own characteristics and different requirements for being stored and processed. First, with production and transaction data, companies choose the infrastructure with the highest level of performance and reliability.
Second, massive video data calls for an economical solution at scale. Finally, edge data storage and processing is a very specific type that requires integrated operations, maintenance and delivery capabilities. As we can see, there are several scenarios, and policies in each are different from another. That is why we need different technologies for such approaches.
In a world rapidly moving into a data-driven economy, what steps do enterprises need to take to cope with the flood of data?
We are in the middle of the Fourth Industrial revolution, an intelligent revolution, which will drive a transition to the digital economy. In such an intelligent world and digital economy, in addition to traditional production factors such as land, labour and capital, data are the decisive assets. We can use oil as an analogy. It fuels all industries starting from factories to retail, and it gives birth to a wide variety of chemicals that we use in our daily lives. Data in the digital economy will be even more impactful. However, less than 2% of the generated data is stored, and less than 10% of the stored data is utilised.
There are two steps enterprises need to take to stay competitive. The first one is to start keeping all the data that is generated in a company’s value chain. The second is to guarantee that data can flow across an organisation. Huawei hopes to work with enterprises to build data infrastructure to solve these two problems.
How is Huawei converging data storage, processing and analysis to create an intelligent data infrastructure?
Traditionally, enterprises stored and analysed data separately. This worked until recently, when agility and speed put pressure on existing infrastructure. These two layers converged to form what we call intelligent data infrastructure. Let me share with you several aspects of it. First, in the new architecture there are fewer data migrations between storage and data processing. The two processes are managed as one and done on the same set of hardware. Second, to accommodate the surging volume of information in an environment where data accountability becomes a must, there should be a level of artificial intelligence (AI). It allows an organisation to make sense of data and improves the performance of the system.
In addition, such a platform should be open, so companies are able to fit the converged system into their existing structure by connecting to traditional IT assets. As you can see, it is not only about the platform itself but also about integrating into the existing environment. Huawei has always adhered to a ‘platform + ecosystem’ strategy, and the converged data platform is no exception. Through technology openness, business cooperation and industry co-creation, Huawei joins hands with partners and customers to build a 5G era data infrastructure with higher performance, better reliability, lower cost and unmanned full lifecycle management.
What is Huawei’s outlook for the expanding all-flash market?
The benefits that all-flash storage can offer to customers are huge. It is faster and more cost-effective and stable than traditional mechanical disk-based arrays. Hang Seng’s electronic securities trading centre improved the number of supported transactions from 60,000 to 120,000 per second after the adoption of Huawei's OceanStor all-flash storage solution.
While some enterprises were early in adopting flash systems, many organisations are still using hard disks. By doing so, they block themselves from new business opportunities that new-generation storage can bring. Huawei has invested heavily in all-flash arrays because the technology is superior to traditional ones.
Just to give you an idea, out of eight storage controllers that we have in our system, seven can fail and still ensure the data is available. We strongly believe that a wide range of organisations will benefit from this move, and that is the reason why we encourage customers to upgrade their storage system to the next level.
As industry data showed, the overall market of all-flash arrays grew more than 14% in 2019. Huawei aims to increase sales in this domain far beyond the industry average, reaching at least 50% growth this year.
As data volumes grow, the move to highly scalable distributed storage systems is on the rise. What is Huawei’s view on this segment?
There is a clear understanding that distributed storage should be scalable and cost-effective. On the one hand, it should meet the requirements of keeping surging amounts of information within the same system. On the other hand, the solution should be economical so enterprises can focus on value creation for their customers. In this regard, we developed the Pacific capacity node with the industry’s highest density, allowing customers to drop the cost per gigabyte they store by 20%. Savings can also come from the separation of storage and computing layers for big data. A leading internet company from Singapore reduced the total cost of ownership by 40% by adopting Huawei’s separation of storage and computing solution.
Can you explain Huawei’s approach to edge data infrastructure?
In 5G era, up to 0.1ms ultra-low latency real-time processing and 99.9999% ultra-high reliability become the key requirements.
Data storage and processing needs to be closer to the physical location where data is generated. Edge data infrastructure becomes more important. There will be exponential growth of edge deployments with the total market reaching billions of dollars in just a few years. Huawei FusionCube is a one-stop delivery solution that is simple and intelligent. It supports out-of-the-box data computing and storage capabilities.
How is automated data lifecycle management changing the way organisations operate? What is Huawei's approach to this?
With a data management system (DMS) organisations can have full control over their data infrastructure with minimal human intervention. A DMS includes capabilities such as planning, rollout, maintenance and optimisation of the data infrastructure, helping customers to reduce opex and improve reliability.
Our goal is to realise unmanned management of the data infrastructure, implemented step by step. Look at autonomous driving, which has five levels of automation. At level three the car can take over most aspects of driving, but the driver must remain alert and ready to take control. With DMS we want to achieve the similar level of automation for data infrastructure.
What current 5G solutions are you working on?
Thanks for asking. There are three 5G-era flagship solutions that everyone will see announced. First, we will introduce a new capacity node for our OceanStor distributed storage solution. With the Pacific capacity node enterprises can move into the 5G-era with confidence that the cost of storage infrastructure is under control as the volume of data dramatically goes up.
We also will launch a new OceanStor all-flash storage solution. We are at least one year ahead of the competition in terms of performance and reliability. Huawei’s all-flash arrays outperform other vendors by three times. We added AI chips in our array to improve efficiency even further.
Finally, we also have an edge data solution, the FusionCube 500. It is a rack with pre-installed components so that enterprises can get it running in the shortest time. We designed it with various industrial scenarios in mind so that it can work in harsh conditions. Now customers do not need to spend time to plan sites, and after FusionCube is deployed, it will take care of itself.
Huawei is working on so much more to see. Let's take on the 5G-era opportunities.
