The hottest wave Peng Zhen focuses on AI and compr

2022-08-16
  • Detail

Pengzhen of Inspur: focus on AI to comprehensively upgrade the layout of smart computing business

on April 16, Inspur cloud data center Partner Conference ipf19 was held in Shanghai, and more than 2000 ISV, Si and other partners attended the conference. At the conference, Inspur systematically expounded the business strategy, product technology layout and new partner strategy of smart computing. Peng Zhen, vice president of Inspur Group, pointed out in his speech that the intelligent transformation of traditional industries is the biggest driving force for market growth. In 2019, the business layout will be upgraded around the typical scenario of intelligent computing "cloud computing, big data and artificial intelligence" to accelerate the goal of promoting the world's first

Inspur server's 2018 - annual market share stabilized in the world's top three

according to Gartner data, Inspur server's sales and sales growth in 2018 were the highest in the world after long-term development, and its share remained in the top three in the world, and remained the world's first in the field of multi node cloud servers

AI business continues to lead the world. IDC data shows that China's AI infrastructure market continued to maintain a strong growth momentum in the first half of 2018, with sales of $543 million. Inspur AI server ranked first in the Chinese market with an absolute advantage of 51.4%, exceeding the total share of all other manufacturers

the storage business is growing rapidly. IDC data shows that in 2018, Inspur storage shipments increased by 27% year-on-year, with the growth rate ranking first in China. It was the first time to be shortlisted in Gartner distributed storage Magic Quadrant and Gartner distributed storage key capability report

the ecological construction of Inspur's partners is also fruitful. By the end of 2018, Inspur had more than 9000 partners, and more than 80 of the top 100 AI companies had established cooperative relations with Inspur. The business of partners increased by 116%, the distribution partners increased by 104%, and there were more than seven distribution partners with more than 100 million yuan. More than 400 joint solutions have been developed with partners, including more than 20 joint artificial intelligence solutions, resulting in sales of more than 20 billion

smart computing focuses on the core of "cloud, number and intelligence" and increases artificial intelligence.

this is the era of smart computing. Wang Endong, academician of the Chinese Academy of engineering and executive president of Inspur Group, proposed at the conference that in the era of wisdom, computing power is productivity. Smart computing has profoundly affected all industries and even promoted the business model innovation of industry economy. In the telecommunications industry, 5g promotes the large-scale application of SDN and nfvi, and promotes the deep integration of ICT; In the financial industry, the combination of artificial intelligence, Internet and finance has changed the financial format. Information technology is no longer an electronic tool to improve efficiency, but a foundation for financial innovation and inclusive financial services; In the automotive industry, the combination of new energy and smart computing will transform the automotive industry from a mechanical industry to an electronic industry, and each vehicle will be a data and service terminal in the future; In the medical industry, the combination of artificial intelligence and medical treatment makes precision medicine possible. In addition, the combination of manufacturing, government, transportation, power and other industries with cloud, big data and artificial intelligence has greatly improved efficiency and created new businesses

Inspur first proposed the definition of intelligent computing, which takes cloud computing as the platform, big data as the cognitive method, and artificial intelligence as the optimization means. In 2019, Inspur will focus on three typical forms of smart computing, namely "cloud, digital and smart", and invest in artificial intelligence to upgrade the layout of smart computing business

integrate into the open ecosystem and meet the challenge of hybrid cloud

public cloud is mainly built and operated by CSP, traditional telecom operators and other users, and most of them are deployed in large-scale and super large-scale data centers. Therefore, public cloud and large-scale and super large-scale data centers have been developing synchronously. By the end of 2018, there were 430 super large-scale data centers in the world, and each data center can accommodate 50000-100000 servers. The overall capacity is equivalent to times of the global server shipments in 2018. The standardization and scale of innovative technologies in large-scale and super large-scale data centers have promoted the emergence and development of OCP, odcc, open19 and other open computing communities, and made open computing an important future industrial trend

the whole cabinet is the most successful project of open computing and the main form of cloud infrastructure in the future. Inspur is the only common member of the three open computing organizations and has played an important role in the development of three types of standard complete cabinet servers. So far, Inspur has occupied more than 60% of the market share in odcc Scorpio complete cabinet servers

Peng Zhen believes that cloud computing is maturing. The sign of maturity is the rise of hybrid cloud. More and more users begin to relocate ERP, database and other traditional applications from public cloud to private cloud environment, and try the hybrid cloud model, which brings new challenges to the public cloud

at the meeting, Langchao and Intel jointly released the first global reference design of high-density four-way server for cloud computing scenario optimization, crane mount (nf8260m5), which has been contributed to the OCP community. This reference standard will be used for reference by more server manufacturers. This product increases the number of single 2U platform cores to 112 and 48 memory sockets, providing higher deployment density and lower overall cost. It can also bear traditional applications such as databases, making the public cloud more suitable for the environment carrying traditional enterprise applications and hybrid clouds

release cloud platform 5.6, strengthen the ability of full stack cloud

more and more enterprises begin to deploy hybrid cloud. Peng Zhen said that hybrid cloud should support public cloud, also support the connection and scheduling of public cloud and private cloud, and also ensure the compatibility of existing different cloud systems. The technology is complex, so many users have taken a compromise approach, that is, decoupling IAAs, PAAS, SaaS and DAAS layer by layer, and splitting them into different platforms, Different manufacturers provide solutions, deploy and maintain them respectively

the cloud platform 5.6 launched by Inspur is a full stack product with IAAs and PAAS service capabilities. At the bottom of IAAs, it integrates the inner sphere and the inner storagethe inner network to realize the full software definition. On top of that, it realizes the deployment management of openstack and container through the inner openstack and inner k8s. The inner manager provides a unified management interface, which can support the interconnection of different virtualization platforms and private clouds, It supports heterogeneous clouds and the interconnection of multi data center clouds. The second layer is the PAAS layer, which integrates cloud IOP and cloud insight HD, and provides services such as database, middleware, big data, etc. to facilitate users to sort out and integrate applications. No matter how many cloud platforms a user has or how many types of cloud platforms, after deploying the cloud platform, it can logically become a cloud

edge server, specially designed for 5g

if 3G and 4G are the era of cloud computing, then 5g is the era of edge computing. According to the definition of ITU of the International Telecommunication Union, the three scenarios of 5g are embB enhanced mobile broadband, urllc ultra-high reliable ultra-low delay communication and mmtc massive Dalian central and local environmental protection policies. 5g communication network needs to deploy small-scale or portable data centers at the edge of the network to localize terminal requests to meet the ultra-low delay requirements of urllc and mmtc

Inspur has launched two servers ne5260m5 and ne5250m5 specially designed for edge computing. The two products meet various standards in the fields of servers and telecommunications, can withstand extreme environments such as high temperature and humidity, and can be suspended for deployment. The ne5250m5 is specially designed for edge AI computing and can be maintained before use. The ne5260m5 is the first general-purpose edge server that meets the otii standard. It supports two Intel cascade Lake Xeon scalable processors, 16 DIMMs, 6 PCI-E slots, 6 2.5-inch hard disks, and 2 m.2 SSDs, which can meet the needs of video surveillance, intelligent manufacturing, IOT and other edge computing applications

big data solutions, released by self-developed nvme SSD and the new generation G5 full flash storage

Peng Zhen said that almost every enterprise is facing data challenges and hopes to turn data into the core competitiveness of the business, but most enterprises still haven't found the appropriate technical route and technical solution. The data innovation framework can be divided into four levels: data source, storage and reading, analysis and mining, and value innovation. Inspur focuses on two levels: data storage and reading and data mining

data can be divided into two categories: big data and small data. Most of the small data are high-frequency read-write data, such as relational data of databases, data warehouses and other systems. After Inspur released as13000g5, a super large-scale data center level distributed storage platform, in March, on the ipf19 scene, Inspur also released a new generation of full flash storage G5, including mid-range hf5000g5, mid-range and high-end hf6000g5, and high-end hf18000g5. High end hf18000g5 better supports the application of massive small data and more agile response with the ultimate performance of 15million IOPs and 0.2 millisecond delay. Inspur also released its independently developed nvme SSD solid state disk

the combination of new generation G5 storage and Yunhai insight HD can provide users with an overall big data solution, and Yunhai insight is almost blank in most places; On the other hand, China's industry has developed rapidly in recent years. T HD V4 is an enterprise level big data basic platform software developed by Inspur based on Hadoop open source framework. It has leading advantages in large-scale cluster one click deployment, data intelligent analysis and mining, heterogeneous platform hybrid deployment and resource refined management, multi data source collaborative computing and three-dimensional data security, and has the super large-scale deployment capability of more than 10000 nodes, It can undertake the collection and access, diversified storage, scale calculation and analysis and mining of massive data of enterprises, and support the rapid implementation of business models of enterprise data centers

according to IDC data, in the first half of 2018, the sales volume and sales volume of China's AI infrastructure market, including 12 fields such as finance and telecommunications, increased by 293% and 219% year-on-year respectively, far higher than the market level. The traditional industry market is about to replace the Internet as the main driving force for the growth of the AI market. Previously, IDC, Accenture and other research companies have expressed the view that the industry market is the next larger development space of the artificial intelligence industry, with a scale of trillion, which is becoming a reality

be a leader in AI

Inspur is a global leading AI computing enterprise and one of the few manufacturers with the ability of AI full stack solutions. It is committed to building an agile, efficient and optimized AI infrastructure at the four levels of computing platform, management suite, framework optimization and application acceleration. In 2018, Inspur launched the artificial intelligence super server agx-5, with a performance of 200 billion times per second. It is the industry's single machine performance strongest artificial intelligence computing system. The management software aistation platform software developed by Inspur is widely used by customers to help Dahua build an artificial intelligence environment at the second level, and realize the intelligent scheduling of GPU resources. The resource utilization rate is increased by 75% and the performance utilization rate is increased by 100%

in addition, Inspur began to participate in and even dominate the product technical standards and performance evaluation standards of artificial intelligence. In the OCP community, Inspur joined the OAM (ocpaccelerator module) project team and participated in the formulation of the first generation OAM standard. In the global system performance evaluation standard organization spec, Langchao took the lead in initiating the establishment of the machine learning technical committee, which contributed its first workload. Currently, it is the chairman of the first committee. Intel is the Secretary General unit, and cooperates with 12 members, including Alibaba, AMD, arm, HPE, IBM, to jointly formulate the performance test of deep learning, which may lead to the short circuit specification of circuit boards and carry out relevant benchmark testing

create an AI ecosystem integrating openness and agility

at the conference, Inspur released the "Inspur metabrain", which is open to output a full stack of AI capabilities. Inspur metabrain system includes super artificial computing system and agile artificial intelligence

Copyright © 2011 JIN SHI