The informatization reform of the industrial and commercial registration system generally comprises development of enterprise credit information disclosure system, optimization of present operating system and development of electronic business license system. Enterprise credit information disclosure system comprises the national enterprise credit information disclosure system and the various provincial and ministerial disclosure systems. Optimization of present enterprise registration, supervision and management system comprises renovation of operating system, total-process electronic renovation, renovation of data center and renovation of public service network. Development of electronic business license system comprises development of the ministerial electronic business license system and development of various provincial and municipal electronic business license systems based on the ministerial root system as the source of trust.
Hubei Provincial Administration of industry and commerce (hereinafter referred to as the Administration) responds to calls of the State Council and the State Administration of Industry and Commerce by launching the credit information disclosure system and the electronic business license system.
How to store big data: conventional data storage mostly comprises conventional relational database plus collective storage. Such a conventional architecture will generate significant pressures on the network and the storage and finally create bottlenecks in data storage while facing big data storage. Meanwhile, a conventional relational database has inborn defects in the storage of unstructured data. With the continuous growth of the client’s business data, it is necessary to readily expand the storage of data. However, such conventional data storage architecture can’t effectively support the linear expansion of nodes. With the expansion of nodes, performance bottlenecks will occur.
How to make rapid searches among big data: massive unstructured data are generated other than structured data in the era of big data, e.g. texts, photos, videos, etc. But a conventional relational database has inborn defects in terms of massive half-structured and unstructured data searches while the client has massive half-structured and unstructured data during development of credit disclosure system and business license system, e.g. system logs, annual statements, electronic business licenses, etc. Therefore, it becomes a difficult informatization problem how to effectively search for half-structured and unstructured data.
How to respond rapidly to needs for high concurrency: according to the development plan of the State Administration of Industry and Commerce, corporate users will declare and search for corporate annual statements within a certain period of time after the credit disclosure system is formally launched, hence leading to highly concurrent accesses to the credit disclosure system. In face of such highly concurrent accesses, a conventional data processing architecture which mostly comprises high-performance server and centralized storage may respond slowly and even suffer a system breakdown. Therefore, it is a major challenge for the client’s informatization work to choose the right type of data processing architecture.
In the informatization program of Hubei Provincial Administration of Industry and Commerce, Inspur offers a general solution featuring Inspur In-Cloud big data all-in-one machine. This solution provides a computing capacity based on 160 cores and 512G memory and a 300T storage capacity to satisfy the business growth needs of the client in the next three years.
In order to address the bottlenecks of conventional data processing architecture in the scene of big data, the solution that Inspur offers, i.e. In-Cloud big data all-in-one machine includes a brand-new share-nothing architecture that overcomes the bottleneck of the conventional architecture in linear expansion. This solution’s architecture is as follows:
It shows that Inspur’s big data platform is established on the basis of capture of corporate logs, annual statements, qualification data and electronic license which are then efficiently processed by Inspur’s ETL tool before they are input into Inspur’s big data platform. After being processed through advanced distributive technology, these data are provided to various upper-layered application systems of the enterprise.
In a scene where structured, half-structured and unstructured data coexist, Inspur In-Cloud big data all-in-one machine solution offers the client with a distributive file system and a distributive database. For the purpose of data storage, such as share-nothing architecture allows concurrent data storage at multiple nodes and hence creates a very thigh data throughput and substantially increases the data storage speed. The distributive database of the all-in-one machine is a column-store non-relational database which can substantially increase the search efficiency of complicated data.
In terms of data safety, Inspur In-Cloud big data all-in-one machine has a three-copy redundant storage mechanism for data storage. Three data copies are created and kept on three different data nodes. When a certain node accidentally breaks down, the data of the other two nodes still remain available. Such a three-copy redundant storage mechanism not only ensures data safety, but also allows simultaneous access as they are all active copies.
In the informatization program of Hubei Provincial Administration of Industry and Commerce, the conventional data processing architecture can no longer satisfy its application needs as its operating system generates complicated data and requires rapid response to highly concurrent data service needs.
After comparisons, the Administration finally decides to adopt Inspur’s In-Cloud big data all-in-one machine solution as the mainstay of its credit disclosure system.
Inspur’s In-Cloud big data all-in-one machine solution has share-nothing architecture. With the increasing number of nodes, the computer’s general performances improve linearly to guarantee the subsequent linear expansion of the client’s system in the most favorable way.
Inspur’s In-Cloud big data all-in-one machine solution provides the client with a distributive database and search is permitted through key/value pair. Second-level search is realized in a real sense when a corporate user performs fuzzy or accurate searches via name and registration number.
Inspur’s In-Cloud big data all-in-one machine solution has share-nothing architecture and a three-copy redundant storage mechanism. In the scene of high concurrency, simultaneous operations and accesses to multiple data copies are permitted at multiple nodes, hence effectively addressing the problem of slow responses in a scene of high concurrency with a conventional architecture.