Cloud Computing and AI Empower Huawei’s Digital Transformation
In 2018, Huawei ranked No. 72 in the Fortune Global 500 with its global sales revenue reaching USD 94.3 billion (CNY 603.6 billion), a year-on-year increase of 15.7 percent. Currently, Huawei has about 180,000 employees, millions of partners, and operates over 900 branch offices in more than 170 countries and regions around the world.
A number of challenges were overcome to achieve its current status in the industry.
• Lower R&D efficiency: Huawei has a large number of mobile phone models and the process of compiling their Operating Systems (OS) was isolated. In total, the Android software system for Huawei involves about 100 million lines of core code, with some code reused across phone models. The effort to isolate OS development has required an inordinate amount of Research and Development (R&D), which impacted R&D efficiency.
• Complex process: Huawei’s team previously displayed project delivery status on a wall and checked nightly; which was very inefficient. Later, Huawei enabled online delivery services, only to encounter a new set of problems. To complete all delivery tasks, it was necessary for operating staff to log in to more than 20 Information Technology (IT) platforms, which led to many complaints from project partners.
• Ineffective management: In 2014, Huawei’s Consistency of Inventory Accounts and Goods (CIAG) rate was about 78 percent. Inconsistencies detected between accounts and inventory at hand were calculated to cost Huawei over USD 9.38 billion (CNY 60 billion) due to time and manpower inefficiencies of conducting logistical and supply-chain management processes manually.
Many large enterprises around the world still face these same challenges, and year after year Huawei continues to achieve new levels of innovation and operations
agility through digital transformation while maintaining double-digit growth in sales revenue.
Do digital transformation.
Over the past 30 years, Huawei supported business development through the continuous implementation of IT-based transformations. Given the increasingly globalized nature, greater complexity, and increased uncertainty across the company’s lines of business, Huawei decision makers concluded that a centralized management system could not effectively respond to new challenges or realize Huawei’s vision of ‘bringing digital to every person, home and organization for a fully connected, intelligent world.’
Huawei believes that in the future, the model of elite teams operating under full authorization of frontline offices will be adopted. Therefore, the management architectures, operation processes, and IT systems of the entire organization will be changed to reflect this philosophy.
Against this backdrop, Huawei has proposed the following digital transformation objectives: In the next three to five years, Huawei will take the lead in building a digital enterprise environment (Digital Huawei) that implements elite team-centric operations supported through big platforms.
• Externally, Huawei will align with customer business goals and digitize transaction processes to provide Real-time, On-demand, All-online, DIY, and Social (ROADS) experiences for customers, consumers, partners, suppliers, and employees.
• Internally, Huawei will make each business domain digital and service-based, remove information barriers across domains, and achieve industry-leading operational efficiency.
Huawei’s digital transformation equips enterprises with efficient operations for better development. Currently, Huawei has achieved digital transformation in nine core business domains, including R&D, sales, delivery, and logistics — and we hope our experience can light the digital paths for other enterprises, including partners and customers.
•Global Collaboration Through Cloud R&D
R&D is a critical business activity for Huawei. In the past decade, the company has invested more than USD 61.6 billion (CNY 394 billion) in R&D alone. At present, Huawei has 14 R&D centers, 36 joint innovation centers, 1,500 labs, and more than 80,000 R&D personnel around the world.
Traditionally, product R&D covers multiple aspects such as processes, tools, data, and compilation environments. For example, if engineers from global research centers (such as the Beijing Research Center, the European Research Center, and Huawei Technologies India Private Limited) jointly develop a high-end network product, they need to perform joint commissioning, apply for equipment to be requisitioned, build IT environments, and scout locations for new facilities. The entire preparation process can take a month or even several months due to siloed lab environments. Today, with the introduction of the Huawei R&D Cloud, new lab environments are streamlined and can be raised in the matter of days.
Hosted on the Huawei Cloud, the company’s current R&D Cloud decouples engineering development phases based on services. The emulation cloud, continuous integration cloud, design cloud, desktop cloud, antivirus cloud, test cloud, and analysis cloud are provided for R&D personnel. With 100,000 desktop clouds, Huawei has laid a solid foundation for comprehensive cloud-based R&D. The R&D process has been rebuilt to improve efficiency, including the management and allocation of millions of Virtual Machines (VMs) in a centralized manner. The result achieves a minute-level preparation environment and a 2.5 times rate of improvement for resource reuse.
Huawei has achieved a 50 percent reduction of operating cycle times and accelerated product launches by using collaborative R&D cloud operations across regions and globally — and have also rebuilt the R&D security architecture by isolating work groups both on and off the cloud.
Huawei also eliminated the isolation problem when compiling mobile phone OSs. Based on the R&D Cloud, engineers can share a public platform that reduces the time to compile all instances of Huawei mobile-phone software around the globe from hours to minutes.
By leveraging Software-Defined Networking (SDN) and cloud computing, we have flexible scheduling and the ability to quickly build R&D Lab-as-a-Service (LaaS) environments.
•Sales Operations Using Big Platform Capabilities
Compared with other services, sales often experience greater uncertainty and more digital challenges. In the past, after detecting slow business growth, frontline directors could not quickly determine whether the lack of growth was caused by errors with lead generation, failure to fulfill project success goals, or problems with key account projects.
Huawei now delivers different user experiences based on specific roles within a sales team with the construction of a large support platform that integrates collaborative spaces, project management services, knowledge platforms, and expert resource platforms. For example, the command system provides a visualized digital experience for managers; the operations system allows project managers to operate sales projects both online and offline; and the automation system provides online automatic and intelligent operation experiences for contract and commercial personnel — the results are improved team collaboration, greater business operations efficiency, and a higher quality frontline user experience.
At present, sales managers at all levels can view real-time operation status for more than 200 subsidiaries across all 170 countries using their mobile phones. The effect has been a gradual increase in the efficiency of a worldwide, collaborative sales operation.
•Real-time and Visualized Integrated Service Delivery
Huawei executives, customers, and partners agree that digital transformation has brought the biggest changes to a complex delivery services process that involves implementation, acceptance, project management, outsourcing management, and resource management.
By integrating outsourced management, site acceptance, receiving, technical support, and resource and personnel management into one operations platform, Huawei has built a one-stop delivery platform plus a series of IT equipment under a service-oriented architecture for delivery personnel. This helps apply online, real-time, visualized, and efficient frontline service delivery. For example, Huawei’s Xi’an delivery command center uses large screens to greatly improve service delivery efficiency by displaying the status of global delivery project execution at each site.
Huawei delivers to millions of sites each year and each site has hundreds of check items. Manual review would be time- and labor-consuming, so Huawei turns to intelligent machines that take less than one minute to inspect items like antennas, ground cables, and Baseband Units (BBU) — and at present, these scenario-specific intelligent machine audits have been verified and widely promoted worldwide.
Digital delivery cannot be achieved without partner and customer support. For example, in Indonesia, Huawei delivers 300 sites on average by more than 1,100 construction teams using more than 300 shipment vehicles every day, covering more than 1,000 islands.
•Global Manufacturing Operations and Command Centers
Production plan accuracy is often the most difficult control item for any manufacturer, including Huawei, because — as with fresh seafood, mobile phones have a short shelf life, and an incorrect manufacturing plan that produces millions of unsold phones is always a disastrous result.
In Huawei’s global manufacturing Operations and Command Center, the company has integrated global supplier status information and market requirements in a service-oriented manner.
Huawei built a real-time decision-making system for each business scenario, which supports supplier material warnings, big data quality warnings, predictive maintenance, intelligent cloud diagnosis, software cloud management, and test network security monitoring.
For example, such corrective warnings can be dispatched to the manufacturing process in a timely manner if quality issues are detected during product testing. The system can manage batch replacement and the quality of materials through big data analysis.
ABB robots are successfully connected to Huawei’s eLTE-U solution at the company’s manufacturing base at Songshan Lake in Donguan, China, and their status and alarm data can be sent back to headquarter in real time, providing big data for preventive production-equipment maintenance.
First Pass Yield (FPY) is an indicator of a production line’s product quality. Due to complex business scenarios and large data volumes (traditional business scenarios were tailored to process-level real-time calculations), the FPY calculations for Huawei device manufacturing were consuming excessive amounts of time. High-performance computing platforms enable the FPY to be calculated in real time (using data integration, cleaning, and display), and manufacturing operations monitoring can be visualized.
In the future, the manufacturing industry will embrace intelligent decision-making with the assistance of high-performance computing and Artificial Intelligence (AI).
•Smart Logistics and Digital Warehousing
Huawei has four supply centers to support high-value product and parts distribution to more than 170 countries — and in the past it was difficult to obtain up-to-date transfer status information. Huawei has now implemented network data access and visualized inventory management for more than 100 warehouses around the world. And at the same time, Huawei’s overall CIAG rate and asset operations efficiency have been greatly improved.
Huawei’s crucial digital transformation measures for key logistics nodes include:
The smart logistics and digital warehousing project at the supply chain logistics center in Songshan Lake is a good example for gradually increasing profitability. The project is a preliminary implementation for visualizing the entire logistics process, for which a series of products have been developed for sending and receiving route reservations, loading simulations, and adding Radio Frequency Identification (RFID) tagging applications that enable real-time asset tracking.
Key warehouses use Huawei’s broadband and narrowband eLTE wireless communication and IoT platforms to manage logistics automation devices such as Automated Guided Vehicles (AGVs) and automatic scanning machines. Trays, forklifts, and other assets are connected to narrowband IoT networks, and key warehouses have applied functions such as automatic inventory tracking, automatic counting, accurate asset placement and tracking, and fast digital warehousing.
Huawei has actively introduced AI into internal and external logistics activities to support the company’s production base in Shenzhen to deliver large numbers of products to project sites worldwide every day.
High utilization packing-space rates can greatly reduce operating costs, and Huawei has successfully increased the packing rate by 8 percent by using an intelligent packing algorithm. The company is committed to improving volume-estimation accuracy through machine learning where models refer to overall configuration information from historical shipment data, adopt basic tally and packaging rules, and create pre-warning alerts for new codes. These efforts have increased shipment accuracy from 40 percent to 80 percent, and quotation accuracy from 30 percent to 70 percent.
Currently, Huawei has introduced Cloud Enterprise Intelligence (EI) to logistics and warehousing operations to further improve efficiency and reduce costs. The results show that EI technologies are improving sorting and packing efficiency by 20 percent and reducing exceptional costs by 30 percent through logistics path-planning.
•Quick Settlements and Automatic Payment
Huawei has over 200 subsidiaries that operate in more than 130 remote offices around the world. Administratively this requires legal compliance with three different accounting standards (China, international, and local) for each office. In the past, the subsidiary accounting process was long and complex.
Huawei can now monitor global closings, including the ability to track and manage complex processes by adopting the following measures:
• Transaction accounting automation.
• Enterprise Resource Planning (ERP) optimization, including cross-Set Of Books (SOB) closings, cross-SOB entry book generation, and fewer manual account closings.
• Data scheduling optimization (one-day, multi-time schedules plus automatic periodic schedules).
• Data quality monitoring (pre-transaction data quality check).
• Improvements to data-analysis-platform performance.
Payments are an important part of finance, and Huawei has a wide spectrum of businesses and a large amount of data in over 70 systems that require staggered payment times and different sums. In the past, the entire payment process employed a six-layer management mechanism — including invoice receipt and scanning, bookkeeping by the accountants responsible for preparing notes, bookkeeping by the review accountants, bookkeeping by the payment accountants, audits, payments, and daily reconciliations — that was processed by manual comparison, which caused low efficiency and frequent errors.
Using a high-performance computing platform, Huawei has shifted from manual to automatic payment processing, and in so doing has improved the security and efficiency of the accounts payable department. Automatic payments have reduced the error rate to 0.0032 percent, shortened IT system convolutional calculation time, and reduced historical billing data monitoring time from nearly 3 hours to 5 minutes.
•Fully Connected Collaborative Offices
Currently, about 180,000 employees of Huawei are distributed among more than 1,000 offices around the world, and their daily work involves interaction with various types of application software. In the early stages of mobile offices, multiple-siloed mobile Apps were constructed to quickly meet service requirements. Incomplete applications integration resulted in an inconsistent user experience that required multiple, independent Apps for instant communication, email, information bulletins, document processing, and service approvals. Further, communication with customers and partners had to be conducted via offline conferences, telephones, and emails. The situation was very inefficient for everybody.
To solve the many problems, Huawei built a cloud-based, mobile, and fully connected collaboration platform, WeLink, to connect people and devices with knowledge and services.
WeLink integrates advanced collaborative services and technologies like instant messaging, email, video conferencing, live video, knowledge base, task management, and intelligent devices to improve the overall efficiency of individuals, team collaborations, and cross-region collaborations.
For example, WeLink videoconferencing has been integrated into multiple scenarios such as Huawei’s remote acceptance, remote customer communications, and remote interview sites to empower Huawei employees with real-time internal and external communications. This means employees take fewer business trips, which in turn reduces costs.
WeLink was designed in 2016 and released in early 2017. The system currently logs 120,000 active users in 170 countries every day, and is a great demonstration for a higher level of work collaboration among Huawei employees.
•Huawei Smart Campus
Huawei’s campuses are located in more than 170 countries and regions.
With the need to manage more than 4 million connected objects in more than 170 countries or regions around the world, there was a clear need for Huawei to build a Smart Campus platform for internal use. In the past, all local security, facility management, and other campus subsystems were built independently, which caused the following problems:
• Difficult data integration and application convergence.
• Failure to cope with increasingly complex security events and inability to meet required management policies.
• Out of proportion manpower costs for analysis and processing due to limited applications intelligence.
Moving forward, Huawei is in the process of aggregating 24 separate campus subsystems to establish a global-unified digital operations center for more highly streamlined data that will improve security protection, response speed, and the quality of service experience.
Huawei has built a fully connected digital campus, which includes an ICT infrastructure, a digital enablement platform, and an Intelligent Operations Center (IOC). The digital enablement platform integrates multiple platforms: including video cloud, big data, integrated communications, IoT, and other supporting platforms such as the GIS/BIM and application engines to create a unified database. In this way, the digital enablement platform will operate on the ICT infrastructure to aggregate multi-dimensional data and provide unified interfaces for upper-layer IOC applications.
The Huawei Smart Campus has changed corporate campus service operations from plan-based control to on-demand supply. The effect brings a significant change by matching supply with demand. The goal of the Smart Campus is to fully connect people, objects, and environments to construct secure, smart, and green campuses for different scenarios.
Taking the pilot Smart Campus at Huawei’s Shenzhen HQ training center as an example: Since the project was launched, incident response times have shrunk from an average of 7.5 minutes to 2 minutes, and the incident handling efficiency has shown a 50 percent improvement; energy efficiency was increased by 10 percent; and the lifecycle of terminal devices was extended by 10 percent. Overall, the intelligently organized experiences of employees and visitors on campus have noticeably improved. Currently, this solution has been put into commercial use in Huawei’s three campuses and the success has accelerated the determination of the company to deploy the internal Smart Campus solution worldwide.
•Mobile Phone eCommerce
In 2017, Huawei delivered 153 million smartphones (including the flagship ‘Honor’ series) in part by leveraging online channels that delivered an impressive sales performance. Huawei built an intelligent online transaction system, iDeal, that employed an IT architecture with decoupled front-end, middle-end, and back-end layers for efficient order handling. iDeal has a demonstrated capacity of processing 1 million orders an hour and has recorded sales revenues of USD 10 billion (CNY 64.4 billion) per day. To be specific:
• The front-end system provides a ROADS experience to meet different requirements for Business-to-Business (B2B) services (for carriers and large partners), Business-to-Partners (B2P) services (for distributors), and Business-to-Consumer (B2C) services (for consumers on Vmall, third-party open-platform malls, and other websites), and supports ‘store openings’ within one to two weeks on a third-party open platform.
• iDeal leverages Huawei’s internal capabilities (such as supply and finance) and third-party ecosystem capabilities (such as logistics and payments) to implement intelligent, automatic, and end-to-end transaction processing.
• The back-end system integrates manufacturing, accounting, and auditing to support the implementation of BP/C transactions.
“Those who produce parachutes should try them out first,” said Huawei Rotating CEO Guo Ping at Huawei Connect 2017, in Shanghai.
In keeping with Mr. Guo’s truism, Huawei applies digital transformation internally to experience all of its risks and difficulties first hand, which, in turn, teaches the company how to provide better value services for customers.
Huawei’s nine core business practices have been operationalized by deploying the company’s leading ICT products and solutions — including innovative digital platforms such as cloud data centers, campus networks, network security, video cloud platforms, converged communications, Edge Computing-Internet of Things (EC-IoT), and eLTE broadband wireless access platforms.
•Cloud Data Center
Through innovations in silicon chips, packaged products, and system architectures, Huawei builds and delivers the industry’s most complete cloud data center solution — that provides supreme performance in multiple aspects, such as OceanStor Dorado V3 all-flash storage appliances with 0.5 ms latency and 99.9999 percent availability, and the CloudEngine data center switch that supports fully programmable 100 Gbit/s networks.
Using its distributed data center and network solutions, Huawei has built eight global ‘service circles’ that provide 100 millisecond access to end-users from anywhere in the world.
Following the principle of ‘one egress for one country,’ Huawei has deployed more than 400 local access lines worldwide. Greatly simplified data center networks that achieve automatic network deployment and intelligence using SDN are built using Wavelength Division Multiplexing (WDM) and passive optical components that reduce the number of deployed optical fibers to a fraction of what would have once been required — and with greatly reduced power consumption.
•Video Cloud Platform
Under the design concept of ‘one cloud, one data lake, and one platform,’ Huawei has built an open, shared, and intelligent video platform with cloud computing and a video big data ‘lake’ as the core.
Videos, images, and existing data are aggregated on one platform and shared between multiple service departments. The platform decouples data from departments and enables them to transform from passive response and hour-level innovation, to proactive prevention and real-time innovation. The Huawei Video Cloud can quickly identify an employee based on nearly 20,000 facial images and allow them to enter the campus within one second, just by scanning their faces.
Both Huawei’s core business transformation and differentiated ICT products and solutions are gradually converging cloud computing and AI technologies to make businesses and products smarter. How does Huawei apply the two technologies?
The diversity and complexity of Huawei’s business has required the development of a ‘multi-cloud’ enterprise IT system architecture in order for the company to thrive.
Like many government agencies or other global companies, Huawei came to the decision that building a private cloud was the key to managing growth and securing critical business data.
After more than 20 years of construction, Huawei has Enterprise Resource Planning (ERP) software packages and physical machines that share capabilities for the improvement of IT asset utilization — and has now moved those applications and IT infrastructures into the cloud to the benefit of company business.
First, enterprise cloudification is a gradual process. ERP in the cloud will not transform into something completely new, but will blend into the background as a backbone application. Second, Huawei has actively embraced more than 10 public cloud services and deployed either non-critical services or those that require elastic resources. For example, we have adopted Office 365 in our daily work, and Salesforce for sales services. And on November 11 each year — known in China as Singles Day — Vmall uses Huawei cloud services to meet the resource elasticity requirements for mobile phone sales.
To support digital transformation, Huawei IT systems must have multi-cloud management capabilities, and making IT systems compatible and collaborative with on-premise software package applications and innovative cloud applications has been a long-term challenge for Huawei. The company is addressing these challenges in the following ways:
•First — Follow to the ‘On Premise + Cloud’ Strategy
At present, we believe this strategy is a good choice for protecting investments, ensuring service stability, and completing the cloud transformation of enterprise IT applications. During application cloudification, on-premise software packages will be preserved as the IT-application backbone, play an important role while innovative cloud applications are created and deployed, and on-premise software packages and cloud-based applications will coexist indefinitely.
•Second — Provide Enterprise IT Systems with Multi-Cloud Management Capabilities
Integrating and scheduling cloud services from multiple providers to support businesses through multi-cloud management is an important strategy for Huawei IT cloudification. Our goal is to integrate multi-cloud resources and services, leverage multi-cloud advantages, reduce cloudification costs, provide seamless multi-cloud environments, and ensure information asset security.
Although different from the hybrid cloud strategy that focuses only on the integration of private and public clouds, Huawei’s multi-cloud management strategy also enables connectivity between public clouds and solves problems that emerge when multiple public and private clouds coexist.
The three core capabilities of Huawei IT ‘multi-cloud management’ are: Multi-cloud secure access, multi-cloud service management and agent, and multi-cloud application development and integration. In this way, Huawei can more quickly respond to business requirements, adapt to changes, and achieve rapid business expansion around the world.
Finally, from an overall perspective, cloudification is only a single step. One priority of enterprise IT systems is to improve customer satisfaction, business operation efficiency, and user experiences through results-oriented improvements, and integrate internal and external capabilities to achieve IT-as-a-Service.
The ‘turning point’ of enterprise applications has come, and in recent years we have witnessed significant breakthroughs such as AlphaGo Zero, TensorFlow, and Huawei EI in AI algorithms and computing capabilities.
Enterprises are determined to use big data and AI to solve problems, and currently, many have prioritized data collection and applications. However, large amounts of data are located in silos formed by isolated enterprise applications. Although some enterprises have achieved data interworking and sharing, because their data has not been classified it is not feasible to apply a single policy that is applicable for all of it.
Huawei prepares for the effective use of data with the following approaches: Unify databases and governance policies, build big data analytics, and AI platforms, and provide big data and AI services.
First, Huawei has formulated an overall data governance strategy: Driven by data, Huawei plans unified enterprise big data platforms that manage data on four planes to build a foundation for enterprise digitization.
• Transaction plane: Use relational databases to process contracts, orders, and other transactional data to maximize the advantages of relational databases in business logic and transaction processing.
• Computing plane: Offload high-value transaction data with complex computing logic to high-performance, in-memory computing and analysis hardware to support real-time decision making.
• Analysis plane: Build big data analytics platforms to support large-scale unstructured and semi-structured data processing.
• Search plane: Respond to large-scale structured and unstructured hybrid data acquisitions and high-concurrent queries in a timely manner.
Second, using open-source software technologies, Huawei has built a ‘service middle-end system’ based on big data analytics and AI that includes AI services, AI training and reasoning models, and big data analytics services.
In addition, Huawei has built a company-level, unified database for the central management of enterprise transactions and third-party data, including access to mass ERP data from service centers through Xdata access and conversion.
On top of this, Huawei will build ‘AI-Inside’ full-stack products and solutions to improve competitiveness for cloud, pipe, and device platforms. Huawei’s three-dimensional AI architecture includes full-stack AI deployment and applications that cover chips, algorithms, products, networks, cloud services, and O&M.
The future evolution of digital transformation remains uncertain. However, one thing is clear: Enterprises must actively embrace digital transformation to survive and thrive. According to IDC’s survey on the global top 2,000 enterprises, 67 percent of CEOs worldwide made digital transformation a core strategy in 2017. As of July 2018, 211 enterprises in the Fortune Global 500(with 48 in the top 100) have chosen Huawei to be their digital transformation partner. We expect that more enterprises will collaborate with Huawei on the road to digital transformation and build a smarter world in the future.
In the past, it would be enough if an ‘elephant’ could ‘dance’ while in the digital age, the ‘elephant’ might also need to ‘do hip-hop dance’. We must actively embrace digital transformation because time and tide wait for no person.