Summary

computing Network integration has become an important starting point to accelerate the digital transformation of the industry because it can provide efficient computing power, advanced storage power and high-speed transportation capacity. The evolution of computing network integration is also accompanied by challenges such as multi-cloud collaboration, cloud network Collaboration, Cloud number collaboration, and computing collaboration. At present, the demand for cloud migration in all walks of life is soaring, but most of them choose the hybrid cloudy deployment method. As a data carrier for business, how can Data storage coordinate with the efficient development of data storage and cloudy, it has become more and more important to promote the integration of computing networks. We have an insight into the development trend of data storage in the process of multi-cloud development, as well as the challenges faced by data management and data flow in a multi-cloud environment. At the same time, we have proposed the collaborative development of data storage and cloud native, helps enterprises select suitable storage types in a multi-cloud architecture.

Introduction

since the first commercial use of cloud computing in 2006, with the acceleration of global digital transformation, cloud computing is gradually becoming the preferred digital business platform for economic development. & ldquo; 14th Five-Year Plan & rdquo; Digital economy development plan clearly points out & ldquo; Promote the integration and development of cloud-network collaboration and computing network & rdquo;, & ldquo; 14th Five-Year Plan & rdquo; the National Information planning clearly requires the construction of new computing power infrastructure. According to Gartner's calculation, 2015 & mdash; The penetration rate of the global cloud computing market increased from 4.3% to 15.3% in 2021 [1], and the scale of the global public cloud market reached 30.7 billion US dollars in 2021, the growth rate has reached 32.5%, and users have spread all over the Internet, government affairs, finance, education, manufacturing and other industries. China has become an important cloud computing market in the world. In 2021, the scale of China's cloud computing market reached 2.9 billion yuan, with a year-on-year growth of 54.4%. The public cloud market reached 218.1 billion yuan, with a year-on-year growth of 70.8%, the private cloud market was RMB 104.8 billion, up 28.7% year-on-year [2]. Among them, the Internet is the mainstream application industry of cloud computing. Driven by policies, the scale of government cloud in China continues to grow at a high speed. The application level of cloud computing in finance, transportation, manufacturing and other industries is also rapidly improving, becoming the most potential field.

Although cloud computing is developing rapidly, it is becoming more and more complex. In the past, many enterprises had the vision of putting application loads on a single cloud. However, with the development of business requirements and technologies, hybrid multi-cloud architecture has become more attractive because it can provide enterprises with more choices. Cloud migration has changed from a single cloud architecture to a hybrid multi-cloud architecture. In the multi-cloud architecture deployment mode, how to implement data security management, efficient circulation, and agile changes has become a key issue for enterprise managers when choosing a multi-cloud architecture.

1. Development trend of hybrid multi-cloud and storage technologies

1.1 diversified enterprise demand promotes explosive growth of cloudy market

when an enterprise decides to use cloud computing as its infrastructure, it needs to consider many factors such as time, cost, business innovation, risk control, legal compliance, and protection of key information assets. Especially for large and medium-sized enterprises with global business and diversified business, it is difficult for a single cloud merchant to meet the business needs of wide distribution, high coupling and complex systems. Therefore, multi-cloud architecture has become a new normal. The advantage of multi-cloud is that enterprises can choose the cloud service that best suits the load of each application. They can choose multiple cloud platforms or different cloud forms without being locked by a single cloud resource, it can also meet the requirements of cloud access closer to the data center. According to Flexera statistics, as shown in Figure 1, 89% of enterprises chose the multi-cloud deployment mode, 80% chose hybrid cloud, 7% chose hybrid public cloud, 2% Selected hybrid private cloud [3]. Therefore, the hybrid multi-cloud deployment mode is the choice for most enterprises to migrate to the cloud.

1.2 Development Trend of data center storage technology

with the rapid development of artificial intelligence, big data, 5G and the internet of things, the data scale has shown explosive growth, and data has become a key factor in production. As the direct carrier of data and an important component of the data center, storage protects the lifeline of the development of the digital economy. The China computing power conference released the white paper on China's capacity (2022), which clarified that the capacity of data center capacity was 800 EB and elaborated the concept of data capacity, it is pointed out that the development of data storage technology and the level of data storage capacity are the key factors to measure the high-quality development of digital economy, and are the basis to activate the value of data elements and improve the ability of data aggregation [4].

Gartner's 2022 storage and data protection technology maturity curve report shows that storage development trends mainly include the following key directions: 2021 & mdash; In 2025, the proportion of NVMe-oF technologies used in external flash storage that supports primary storage workloads has increased from 5% to 30%;2022 & mdash; In 2025, the number of storage vendors that have introduced at least one hybrid cloud architecture deployment mode has increased from 15% to 60%;2022 & mdash; In 2025, the proportion of enterprise storage deployed at the edge increased from 15% to 40%. In 2026, in data centers, the unstructured data capacity of public cloud and edge deployment is expected to be 3 times that of 2022 [5]. In the field of cloud computing storage, container storage deployment and storage integrated with proprietary hardware of data processing units have also become two major trends.

2. Challenges to the construction of data infrastructure layer in hybrid multi-cloud

2.1 data management challenges caused by explosive data growth

according to IDC report, 2022 & mdash; In 2026, the data generated by enterprises will increase from 59 ZB to 155 ZB, and the scale of the global data field is expected to more than double. The growth rate of enterprise data will be twice that of consumer data [6]. ESG's multi-cloud maturity survey also shows that the three-year compound annual growth rate of unstructured data managed by enterprises is 39.4%[7]. This will pose greater challenges to enterprise management and data protection, especially in a cloudy deployment mode. In addition to using different types of cloud services such as public cloud, private cloud, and Apsara stack, the multi-cloud environment also uses resource services from different cloud computing providers, in a multi-cloud environment, the biggest problem is how to manage the separated data and resources in a unified manner and deal with the challenges of low data utilization, complex operation and maintenance, and high operating costs.

Seagate report points out that only 32% of the available data of enterprises will be used in 2020 [8]. Due to the different architectures between multiple clouds, there are barriers to data access. Therefore, it is difficult to implement multi-cloud data storage expansion, easy data circulation and utilization, and it is easy to generate new & ldquo; data Island & rdquo;.

Retaining data between different clouds limits the mobility of business loads and increases compliance risks. In particular, the storage Cost of cloud computing is unstable at present. With the exponential growth of data and the increase of Cost, hidden expenses have greatly increased the Total Cost of Ownership,TCO), and the extra high cost will lock the data on a cloud. For example, 73% of the interviewees said that the cost of data retention limited their enterprises' ability to maximize the value of data. Another 60% of the surveyed enterprises said that they deleted business data that they considered valuable due to high storage costs [7]. As you can see, more and more enterprises are experiencing the challenges related to the cost of cloudy data.

2.2 multi-cloud makes different data resources become isolated islands

because multi-cloud construction uses different technology stacks and different data platforms to integrate and manage data, it is difficult for enterprises to migrate data in different formats that reside in different storage technologies. Although enterprises integrate various basic resources such as public cloud, private cloud, traditional enterprise data center and edge data center to support various cloud applications, data centers with different technology systems are still & ldquo; data Island & rdquo;, it is very difficult and costly to share and flow data with each other. High costs of obtaining and maintaining removable solutions and lack of mobility may lead to the risk of cloudy strategies. Due to the long-term existence of & ldquo; Data Island & rdquo;, data migration and circulation face more security risks, how to implement compliance management for data on different clouds is also a huge challenge.

2.3 difficulty in data resource flow between multiple clouds

in cloudy scenarios, data resources often need to be processed by different application platforms, which requires that data can flow freely between different cloud platforms, however, each cloud vendor uses different storage layer hardware resources, which leads to the situation that hardware resources are locked by cloud vendors, enterprises are forced to use different storage hardware resources when building multiple cloud platforms. However, restrictions such as protocol incompatibility, ecosystem incompatibility, data inconsistency, and data unavailability between cloud vendors bring risks and bottlenecks to data flow, therefore, it takes weeks or months to complete data migration and data availability. Data flow between clouds, edges, and clouds requires a large amount of network bandwidth and stable low latency. Therefore, unstable network transmission efficiency also brings bottlenecks to data flow. Judging from the global development of cloud computing and storage industry, cloud and storage have established an open cooperation ecology and standards, which are more forward-looking and collaborative. At the business level, cloud platforms are purchased separately from data storage devices. At the technical level, cloud platforms are compatible with data storage devices. A Unified Data Storage resource pool can support real-time data sharing and flow among multiple clouds to ensure complete data consistency.

3 Suggestions on Promoting the collaborative and high-quality development of multi-cloud and data storage

3.1 deploy multi-data storage and build multi-cloud infrastructure through hierarchical decoupling

data storage has always been the best base for high-value data and has developed with the evolution of data applications. Early data applications were mainly database systems and large host applications. At this time, the external Storage was mainly Storage Area Network (SAN). With the rapid development of the Internet and cloud computing, different users have different data forms, application scenarios, and access methods when using cloud computing, cloud computing needs to meet multiple user needs in various scenarios, and unified storage emerges. On the basis of ensuring performance and reliability, storage scalability is also greatly improved. There are many ways to choose the data storage architecture at the IaaS layer, including the distributed architecture based on server storage and the SAN architecture based on external storage. However, no matter which method is selected, most of the current IaaS layers adopt the hierarchical decoupling networking mode of computing, storage, and network separation, even large public cloud service providers. Especially in the multi-cloud deployment mode dominated by hybrid cloud or private cloud, the external storage devices in the data center have more advantages as a hierarchical decoupling architecture. The data management, data flow and cost control brought by it are more in line with the current situation of cloudy development. In a multi-cloud architecture, diversified storage types should be selected and hierarchical decoupling should be adopted.

3.2 centralized data sharing and storage of cloud native capabilities accelerate enterprise multi-cloud digital transformation

Enterprise business innovation continues to accelerate, and the application launch cycle needs to be further shortened. Embracing container technology has become the choice for most enterprises today. On-cloud development and off-cloud deployment can not only utilize the agility and elasticity of public clouds to reduce development and testing costs, but also use private clouds to ensure data security and sovereignty. Due to the inconsistent development and application architectures and business separation between the cloud and the cloud, the consistent delivery and operation and maintenance of applications cannot be guaranteed, the enterprise Information Technology (IT) architecture needs to be containerized when the on-premises PaaS/SaaS service is pushed to the off-premises enterprise data center. When stateful applications undergo cloud transformation, cloud-native application scenarios require high service agility and flexibility. In many scenarios, containers are expected to start quickly and schedule flexibly. The storage system in the cloud native era not only meets the requirements of performance, stability and reliability, but also meets the requirements of business systems for agility. It can be integrated with business launch through a unified orchestration system, and can be quickly scaled out. First of all, storage supports standard interface specifications to connect with containers. Professional storage is used to replace local disks of servers, which has high reliability and flexible replacement and expansion of computing and storage. On this basis, Network Attached Storage (NAS) is the most suitable Storage for containers. On the one hand, business Systems can be easily mounted and unmounted to NAS file systems to meet the requirements of container auto scaling. On the other hand, NAS has a wide range of enterprise-level features, such as multi-tenancy and quota, it can be used together with application-level multi-tenant and Quality of Service (QoS) requirements of containers. In addition, NAS can unload file system overhead to storage. In large concurrent applications, the performance is not affected by host operating system overhead. With the combination of container standard interfaces and disaster recovery plug-ins, data storage can provide application-level cross-cloud and cross-data center disaster recovery for containers, enable cloud-native critical applications to have the same disaster recovery capability as traditional applications, avoid single node failures, single data center failures, and cluster failures, and achieve the available disaster recovery level of zero data loss. This high-availability mechanism only needs to be implemented through the underlying storage layer, and applications do not need to make any modifications to consider high availability, so that enterprises only focus on the application itself and accelerate enterprise IT to move towards a multi-cloud architecture.

3.3 embrace the multi-cloud architecture and create a secure and reliable storage base in the multi-cloud era

data storage is the key to the advantages and disadvantages of a multi-cloud architecture. We should actively develop the capabilities of data storage and cloud technology to build data hierarchical storage, global file systems, storage containerization, and Diskless data center architectures. Explore technical routes such as hierarchical management of local and cloud hot and cold data, data disaster recovery and backup, and create a highly reliable, high-performance, multi-protocol, and low-cost unified storage. Promote from multiple dimensions such as production, learning, research, and use, and encourage the use of data infrastructure construction and upgrading in multi-cloud construction, by formulating policies, enterprises are encouraged to build a unified data storage resource pool during the process of migrating to the cloud, and standards for interconnection between cloud platforms and data storage are formulated. Accelerate the application pace of flash memory data storage and incorporate it into the enterprise digital transformation work plan. We will increase industrial support and investment in scientific research funds for the new storage architecture, and guide enterprises to combine basic research, technological innovation and market application of the new storage architecture with universities and research institutes. Create a number of demonstration projects represented by new storage architecture applications to accelerate the industrialization and application of related technical products.

4 Conclusion

A major trend of the new computing power network system is the integration of computing networks. Under the background of the integration of computing networks, the development trend of cloud computing must be a rapid change to multi-cloud. For the digital transformation of enterprises, it is particularly important to realize simplified access, seamless flow and unified management of data. Improving the adaptability and flexibility under the multi-cloud strategy, reducing the complexity of multi-cloud data management, and solving data management problems such as multi-cloud data layering, disaster recovery, and migration are the challenges faced by each enterprise. Cloud vendors and storage vendors should actively deploy professional data storage to multi-cloud architectures to help enterprises smoothly evolve data across clouds.