Shulou(Shulou.com)06/01 Report--

Deputy General Manager of Information Technology Department of Xia Yan Yingda Foundation

Graduated from Jilin University computer and Communications Whampoa Beijing Post with 20 years of financial IT experience. He is currently the Deputy General Manager of the Information Technology Department of the British University Fund and the Information Technology consultant of the Securities and Futures Industry of the Securities Regulatory Commission, mainly responsible for the IT information construction management and planning of fund enterprises.

Summary:

This article introduces the background, evaluation process, deployment plan and experience of the overall transformation of the UK Foundation IT infrastructure to a super-converged architecture, especially the relevant concerns and targeted evaluation and scheme design for the small machine plus SAN storage part of the original core production business, hoping to provide reference for the industry.

one。 Background introduction

The original IT architecture of the British University Fund has always adopted the following common models in the industry: minicomputer plus SAN storage to run core database, x86 server plus SAN storage to run other businesses:

Figure 1: diagram of the original IT structure of the UK Foundation

The reason for the initial launch of the super-integration architecture is that the UK University hopes to prepare for disasters based on policy regulation and business needs, and has identified the following core requirements:

The RPO of the solution hopes to be less than 5 minutes according to the requirements of the regulatory unit; the budget is limited and wants the most cost-effective solution; due to the limited staff of the IT department, the disaster recovery solution of the data center is expected to be simple to maintain; the infrastructure performance of the production and disaster preparedness center should support the daily business pressure of the core database system, and disaster preparedness needs to have the same business processing capacity. Disaster recovery computer room plans to rent IDC computer room, the space is limited, it is hoped that the less equipment the better, at the same time to ensure the convenience of subsequent expansion

According to the traditional disaster recovery center scheme, it is necessary to buy another set of storage, as well as the replication software that comes with the storage, and two stores of the same brand and model are needed at the same time. The overall scheme is expensive, and the openness, maintenance and scalability are not good. In addition, there were some other problems in the IT infrastructure at that time, such as: Oracle RAC was also deployed on the IBM minicomputer, and the equipment was obsolete. Considering the equipment risk and the complex problems of operation and maintenance, we hoped to phase out the IBM minicomputer as soon as possible.

Because we have heard of distributed storage with super-converged architecture many years ago, and think that this solution may become a trend and gradually replace the traditional architecture, now if we spend a lot of money on the traditional architecture, it may soon be eliminated, so find SmartX, Lutanli and other manufacturers to introduce it. After communication, it was found that there were indeed many advantages of this architecture, and the maturity of the product and the coverage of application scenarios were greatly improved at that time, so we began to evaluate whether the overall architecture could be upgraded to a super-fusion architecture. and build disaster preparedness in the same city based on this architecture.

II. Scheme evaluation and design of super-integration architecture

Compared with traditional solutions, the advantages of super-fusion architecture are still very obvious, such as:

1. The whole IT architecture and disaster recovery architecture based on x86 servers can be simplified, and the reliability of the whole system can be further improved.

two。 With better scalability, IT teams can meet business resource delivery requirements more quickly

3. The total cost of ownership can be effectively reduced.

Despite the above advantages of super-convergence, the team still has concerns about reliability because it involves the transformation of the company's overall IT infrastructure. Can the super-converged architecture based on distributed storage and virtualization really replace the original minicomputer plus traditional storage? After all, there are not many cases in which the business of small computers is transferred to a super-converged architecture.

In response to the above concerns, we have carried out scheme evaluation and design considerations from several directions:

1. Visited a lot of fund companies, insurance companies and other customers to understand the application situation, and learned that super-integration applications are already very common in production systems.

Through the POC of the super-fusion product, the reliability, fault tolerance, performance and function of the product are fully understood and verified. Make full use of the existing equipment combined with Oracle DG technology to do data synchronization at the database level, that is, for the core business, do a double insurance on the basis of super-converged distributed storage copies. Use virtualization-based CDP products with super-integration to build a disaster recovery center to achieve three insurances.

The final framework of the scheme is as follows:

Figure 2: current IT architecture diagram of the UK University Fund

Third, the application effect of super-fusion architecture.

At present, all production systems of Yingda Fund have been migrated to super-integration platform, including O32 system, valuation system, TA system, direct marketing system, various Oracle databases (investment, TA, valuation, risk control, data center), market, quotation, etc. The application effects of super-integration transformation in the past year are as follows:

The performance of the system is good, and there are no problems in stable operation for more than a year. The new architecture based on super fusion can carry all kinds of services, including the services on the original IBM minicomputer. There are only standard x86 servers and Ethernet in the whole data center, and distributed storage is very intelligent in automatic fault recovery, automatic data equalization and so on. Operators can free themselves from the tedious and inefficient operation and maintenance to carry out more innovative work such as new technology research. Cabinet space savings of 83%, using ultra-converged high-density all-in-one machines, previously reduced from 12 cabinets to 2 cabinets for all hardware equipment in the production infrastructure

Figure 3: comparison of cabinet space before and after the upgrade of British University Fund data center

The network of the entire data center has also been greatly simplified, and the following is the network architecture diagram before and after the upgrade:

Figure 4: British University Fund data center upgrade network architecture chart 5: British University Fund data center upgrade network architecture chart

Get through the layer 2 network architecture of production cluster and disaster recovery cluster, and achieve RPO seconds and RTO minute disaster recovery capabilities through CDP tools. Virtual machines on the two clusters backup each other and switch with one button, greatly reducing the complexity of disaster cutting drills, which can be described later. The system has good scalability and fast online speed, and the benefits of Agile IT have been demonstrated by several recent examples. The first example is that the company is anxious to start the test of quantitative investment, which needs to build a test environment quickly, and quantification requires a high-performance server, because we adopt a new architecture such as super fusion, which quickly supports the company's quantitative business. Another example is that the company is going to launch a new business system quickly recently. The test environment built is equivalent to replicating the production environment. Based on the super-fusion architecture, eight or nine virtual machines will be in place very quickly. IV. Summary and experience

Through the stable operation of the company for more than one year, it is proved that super integration can fully carry all kinds of business, including core production, and its agile features such as easy maintenance, easy expansion and rapid launch are also very much in line with the needs of the company's financial technology transformation. the system has been expanded in the near future and will continue to support the company's business development.

Finally, I would like to say that although the super-integration framework has been relatively mature, the product implementation and services vary greatly from manufacturer to manufacturer. We must choose professional product manufacturers who are really competitive in super-integration products, such as Nutanix and VMware abroad, and SmartX in China. Foreign product brands are definitely stronger, but the price is relatively high, and the response speed of local services may not be timely enough. After comprehensive consideration, we finally chose the domestic professional manufacturer SmartX, the price is relatively more reasonable, the response speed of the service is faster, the current result is still satisfactory.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.