Requirements and Challenges

Core HRSS service data flows among municipal bureaus, service management organizations, terminals, communities, and provincial data centers (including disaster recovery centers) mainly through two channels: an e-Government extranet and the Internet. HRSS access networks also expedite data communication. The comprehensive network collects, manages, converges, and shares unified social security card services through these data streams. A comprehensive management mechanism is also enabled for these data streams to ensure data security and Operation and Maintenance (O&M). To this end, governments can extend multi-level, comprehensive HRSS services to the public and insured people by enabling:

• Core service systems to share data with each other
• Internal HRSS organizations to share and exchange data with external collaborative organizations (such as banks, public security, and taxation)
• Four-level service handling mechanism: province-city-county-street

Service Requirements

Specific types of HRSS data need differentiated data processing requirements. Figure 1-1 uses social security services as an example to demonstrate detailed and customized data processing requirements.

Figure 1-1 Processing requirements of core service data

Service data processing

Processing of core service data refers to the process of generating, updating, modifying, querying, counting, and storing the service data in the HRSS service systems.

• HRSS service data processing requires high reliability, consistency, bulk processing, and quick response to concurrent requests. Part of the core service data (such as real-time settlement data) requires real-time data processing. The query data in the public service zone, which are copies of the core service data, requires low-level real-time processing and unidirectional data transmission.
• The HRSS software platform needs distributed computing for its core service data. Specifically, a provincial data center and municipal sub-centers jointly process data based on the type of service data, and the provincial data center provides the only comprehensive province-wide data view.
• Municipal data must be processed in distributed mode, eliminating the risk of a single point of failure (SPOF). When one point is faulty, the provincial data center or other data center nodes will take over the processing of core service data, ensuring high availability of core services.

Shared data exchange module

Exchanges basic data and public service data with service departments or grassroots units

Manages metadata and public code databases and shares the metadata with service departments

Stores and manages public service data using the data exchange platform, synchronizes data with the extranet, and provides one-stop services for the public

Note that the exchanged data must be consistent with the data in the production zone. Exchanged data does not require strict real-time exchange but, rather, limited real-time exchange based on the data exchange period and delay. In addition, data transmission in the shared data exchange module must be controllable and bidirectional.

Decision-making data counting

Statistical data must be converged, stored, and managed. If decision-making is involved, the data must undergo statistical analysis and querying.

Statistical analysis and decision-making data must be comprehensive and displayed in graphs. Note that decision-making data requires only snapshots based on the design principles of data warehouses and no real-time data counting.

Data disaster recovery

Core services must use application-level DR while other services use data-level DR, according to the data DR requirements of China's Golden Insurance Projects and customized requirements of provincial and municipal information systems.

To address architecture requirements of the HRSS core service software platform, data DR systems must provide diverse backup functions, as well as data processing capabilities or multi-center redundancy survivability.

Service concurrency, data exchange, data query, data counting, and data processing performance during the convergence, quick response, and stability in production and exchange zones are also important.

Customer Challenges

Computing architecture

After data is centralized in upper-level networks, multi-point risks change to one-point risks. One single technical fault on a provincial network affects the service operating throughout the entire province. If the technical fault is a big one, the impact will be disastrous. That is why data centralization represents higher requirements for system reliability and data processing capabilities.

Dispersed data resources

HRSS data is scattered in many organizations such as hospitals, drugstores, and service management organizations.

Centralized data processing

The amount of data processing often reaches its peak between 9:00 a.m. and 11:00 a.m. and 2:00 p.m. to 4:00 p.m. in city centers where HRSS resources and insured groups usually reside. Once data is centralized, the data processing system must be able to handle data with different resources centrally within a specified period. Therefore, the HRSS ICT system must ensure strong scalability and service continuity under tremendous real-time data pressure.

O&M

Most HRSS data processing systems use either the x86 architecture or the Reduced Instruction Set Computing (RISC) architecture. Midrange computers use the latter architecture, and most midrange computers are not open to third-party vendors. Therefore, only midrange computer suppliers can effectively operate and maintain their core components. This inevitably incurs many uncontrollable O&M risks. If any faults beyond customer O&M capabilities occur, customers must turn to the supplier because very few third-party vendors can fix these faults.

TCO

The HRSS industry depends highly on investment and is sensitive to the Total Cost of Ownership (TCO), which includes energy consumption, management, and upgrade expenses. Therefore, operation scalability, architecture openness, and technology-sharing of core components must be carefully planned to simplify IT management and reduce TCO.

Service innovation and decision-making efficiency

With the continuous development and innovation of HRSS services, information systems must have a resilient basic architecture to enable new applications to go online within the shortest-possible time. Additionally, a high-performance and scalable core service data processing platform is necessary to support critical decision-making, such as for social security funds and pensions.

Huawei Solution

Huawei provides an HRSS core service data processing solution that features open, reliable, flexible, and secure data processing to help HRSS customers keep abreast of data centralization and service extension.

This solution includes:

• Highly efficient data computing and storage systems
• Data DR solutions dedicated to mainstream HRSS databases
• Convergence-ready integrated devices.

High-Performance Computing Devices and Speed-Acceleration Components

Figure 2-1 Data computing and speed-acceleration components

Core database servers

Top-level blade servers integrate computing, storage, and network functions for service systems that require higher system performance to process a huge amount of data.

High-performance, high-end subrack servers are deployed in two-node clusters for common service systems.

Include multiple CPUs, cost-efficient memory expansion, hard disk expansion, and acceleration of input/output modules. For example, Peripheral Component Interconnect express (PCIe) and a General Processing Unit (GPU) ensure high computing performance.

Application servers

Highly resilient, high-performance blade servers

Use virtualized deployment and load balancing to form application server clusters

Maximize application processing while maintaining flexible computing capabilities

Bottleneck-Breaking Storage System

The design of Huawei's data processing solution focuses on breaking I/O performance bottlenecks, in order to enhance storage system performance and address data centralization requirements. Figure 2-2 describes this high-performance storage system.

Figure 2-2 High I/O performance HRSS storage system

Uses the SmartCache engine with built-in disk arrays to automatically identify hotspot data and caches the data in the solid-state storage pool, thereby sharply decreasing reading delay.

Forms redundant disk groups based on solid-state and traditional storage systems by using Adaptec Storage Management (ASM), writes data updates into the two storage systems, and reads data only from the solid-state storage system, thereby reducing reading delay and improving data reliability.

Figure 2-3, below, shows how the storage system works.

Figure 2-3 ASM-based storage mirroring

Data DR Solutions Open to Mainstream HRSS Databases

Huawei offers customers a comprehensive data DR solution that covers the application, network, and media layers, as shown in the Figure 2-4 below.

Figure 2-4 Data DR solution

Convergence-Ready Integrated Devices

To keep up with the trend of integrated software and hardware designs, Huawei's HRSS core service data processing solution uses devices that integrate computing, storage, and network functions that address HRSS customer requirements. Figure 2-5 helps you better understand how the integrated devices operate.

Figure 2-5 Integrated HRSS data devices

Solution Highlights

Highlights of Huawei's HRSS core service data processing solution include:

Open, easy-to-use, and highly reliable computing devices

• Support the evolution to the next three generations of Intel® processors and the dynamic expansion of computing nodes
• Enable flexible configurations based on specific service requirements, supports stateless computing and migration of physical machines, and ensures high availability of service-critical applications
• Provide highly reliable redundancy for the power supply (N+1 or N+N), fan, and passive backplane

Field-proven high performance

• Improves I/O performance by 20 times, provides up to one million Input/Output Operations per Second (IOPS), and ensures 500 us access delay that is 5 percent of the access delay of traditional disk arrays
• Passes authentication of third-party organizations based on the TPC-C benchmark, Huawei's Online Transaction Processing (OLTP) services:
• The SmartCache engine with built-in disk arrays supports three times more online concurrent users and reduces ticket-handling delay by 70 percent
• ASM storage mirroring supports four times more online concurrent users and reduces ticket-handling delay by 87 percent.

Comprehensive, highly reliable DR solution covering application, network, and media layers

Customer Benefits

• Open computing architecture addresses data processing requirements of HRSS data centralization
• Meets governments' requirements for easy-to-use information system O&M
• Supports HRSS service innovations and improves decision-making efficiency
• High-performance, highly reliable data processing at a lowered TCO