Since the first year of fintech in 2016, various kinds of intelligent scenarios and intelligent concepts such as intelligent customer service, intelligent trading and intelligent investment research have emerged in the securities fund industry. It seems that the whole industry has entered a lofty intelligent era.

From the author’s point of view, there are some applications that really solve the pain points of the industry, but there are also a lot of over-ambitious and blindly chasing hot behaviors.

Here, the author summarizes and tries to analyze the common intelligent XX in the industry, hoping to bring some enlightenment to the securities fund industry and information technology service institutions in the application of financial technology.

The following topics are initially planned:

  • Intelligent Operation and Maintenance – an easy to overlook scenario;

  • Intelligent customer service – a scenario that is easy to cross the line;

  • Intelligent teaching – a scenario that is easy to underestimate;

  • Intelligent trading – a scenario of pursuit of perfection;

  • Intelligent research – a challenging scenario;

  • Intelligent casting – a scene on the road;

  • Smart financial management – a vision full of good.

Today we’re talking about intelligent operations — a neglected scene. Try to put it in colloquial terms.

I. Examples of typical industry OPERATION and maintenance scenarios

Before I start, I should mention that system development, system testing, project management and so on all have their own hard work (will be introduced in the future), but this article is about intelligent operation and maintenance, so this chapter only introduces the bitter tears of operation and maintenance personnel.

In terms of system operation and maintenance, the securities industry has more system types, higher real-time requirements and more complex overall than the fund industry.

The principles of operation and maintenance are the same. The author mainly describes the typical operation and maintenance scenes in the securities industry.

[pre-market] tension. Check the hardware and environment of the main equipment room and disaster recovery equipment room (coordinated operation between the two centers and the three centers).

The children in charge of system maintenance need to restart the exchange, register companies, Shenzhentong and other gateways connecting the core institutions of each market (Ezstep, Eztrans, FDEP, D-COM, Prop…). , all kinds of business systems, middleware, small stations, inventory clearance, download and load all kinds of market information, receive all kinds of parameters, market data and other data of core institutions, complete a series of operations such as system initialization.

Many of these operations have a clear sequence, connection requirements and time constraints. Generally speaking, all major business systems need to be basically ready by 9:15.

A high level of tension. Monitor the operating status of the equipment room and system, inspect each service system (except in extreme cases, the equipment room is generally not allowed to enter), assist the customer manager to solve customers’ common problems in time, assist the business department to solve problems of various service systems, and make emergency preparations for various systems.

Some also need to assist the business department to monitor all kinds of abnormal transaction behavior. Then the heart pray, hanging quietly waiting for the smooth closing market.

【 After the market closed 】 The high tension finally passed. At this time, it is necessary to pay attention to the system used by the trading management department, the Securities and Finance Department, the self-management department and other departments to ensure the timely generation of various daytime trading statements; Then receive and process all kinds of clearing documents sent by market core institutions such as exchanges and clearing companies, assist the settlement department to complete all kinds of clearing and settlement work or provide technical support.

After that, all kinds of data are submitted to the regulatory body (some institutions are submitted by other departments), and the data backup operation of the day is carried out. After the system operation of all business departments is finished, the operation and maintenance personnel also need to conduct a series of operations to initialize the system to the state of T+1 and run batch operations at night, and then quietly wait for the end of work.

It is not uncommon to stay up all night once in a while when a core organization releases data late or a business data is abnormal.

[Most embarrassing holidays] On the one hand, there are many core organizations in the domestic market, which can not avoid upgrading or reforming the system. However, the characteristics of the industry and the culture of system deployment determine that such upgrading and reforming are most likely to be carried out on weekends or holidays, which also requires the cooperation of all business organizations to conduct joint testing.

Operators itself, on the other hand, dozens or even hundreds of system upgrades, and other critical systems are inevitable to the total change of demand, this kind of basic will also work on holidays, and also basic operations staff annual holiday plans went up, large securities firms have paid leave mechanism has flexible response, small brokerage manpower shortage, hard place.

In addition, the holidays can be harder than usual because tests are often based on production systems, which adds to the job of restoring and validating the system after the test (how many businesses in the past have had the next day’s trading disrupted because the system was not in place after the holiday test).

When the outside world sees the work of operation and maintenance, development and project construction, it is just like watching marathon, 100-meter race and rowing in sports meeting. Operation and maintenance is boring for many people.

When it comes to writing work summary, there is always some embarrassment. Other departments make long speeches, while many operations and maintenance work are done once, but repeated N times. There is a feeling of “one day, but repeated 364 times”. In addition, it is even more difficult to carry out work in case of leaders who do not know or understand. The following dialogue also reflects the helplessness of operation and maintenance:

The person in charge of A securities company, T day feeling: “the system this big half year also have no what action, operation and maintenance of dozens of people are doing?” T+N day, the system broke down, A sighed again: “even the system can not handle, what is the use of raising this group of operation and maintenance people?” .

In short, this type of leader is completely unaware of the value of operations personnel. Of course, operation and maintenance personnel should also strengthen the initiative to communicate with leaders and colleagues.

In fact, if A really think so, the hospital, 120, 999, physical examination center is really not necessary to exist, as long as the delivery room is ok.

In addition, it should be pointed out that the most knowledgeable and blind spots of business process are often senior business system operation and maintenance or support personnel (different organizational structure, some securities companies will also put these people in the business department, similar to the role of SA).

Second, the importance of operation and maintenance

Everyone knows that operations is important. Have you or your team thought about where that is? According to the author, the importance of O&M is mainly reflected by the impact of faults, which includes at least the following:

One is customer influence. The system is interrupted, business continuity is affected, customer transactions are affected, or customers are lost.

The second is reputation. A major system failure will affect the company’s reputation, and reputation is valuable.

Third, the impact of comprehensive operation. This should be the most important impact factor, but also the important characteristics of the system operation and maintenance of the securities fund industry.

According to the “Securities Futures Industry Information Security Incident Report and Investigation Handling Measures” (CSRC announcement [2012]46, referred to as “Announcement 46”), the information security events are divided into four categories: particularly major events, major events, major events and general events.

According to the “Regulation on Classification of Securities Companies (2017 Revision)” published on the website of CSRC, the evaluation indicators and standards of risk management ability of securities companies. It mainly includes capital adequacy, corporate governance and compliance management, total risk management, information system security, customer rights and interests protection, information disclosure and so on.

Information system security mainly includes perfect IT governance, independent and effective information system management mechanism; The information system has complete functions, which can effectively meet the needs of customers, such as entrusting, trading, clearing, account opening and inquiry, and ensure the security of customers’ electronic data and other information. The information system runs safely and stably, avoiding frequent information security accidents or major accidents; Information system emergency plan is effective, can timely respond to information security accidents and other four categories.

A major information security events are likely to affect the level of the classification of the company, thus affecting investor protection fund to pay the amount, also may affect the application of new business qualifications, credit bank loans, bonds brokerage business, insurance fund in the forehead, new business application, the existing business acquisition such as stock pledge, and facing more strict compliance.

As can be seen from the following figure, the classification level does have a greater impact on the payment of securities investors protection fund. According to the relevant notice of the CSRC in 2013, when the size of the protection fund is more than 20 billion, AAA, AA, A, BBB, BB, B, CCC, CC, C, D and other grade 10 securities companies, They pay into the protection fund at the ratio of 0.5%, 0.75%, 1%, 1.5%, 1.75%, 2%, 2.5%, 2.75%, 3% and 3.5% of their operating income respectively, and the relationship between the two ends is seven times.

In conclusion, all the business industry is carried out through the system, all operations are operators IT priority, is the so-called 1, ignore the other IT of operational activity is edge lick blood, operations are easy to be ignored, because we do not feel his presence at ordinary times, once felt, would be a problem, which is similar to the health of human body, Healthy people are unable to feel the existence of the viscera, one day you feel his existence, large probability there is a problem.

3. Complexity of operation and maintenance

The complexity of industry system operation and maintenance is determined by internal and external factors. Internal cause, because industry main business are to obtain licenses or approved only after acquisition, to launch any type of business to business organization system preparation time is shorter, management institutions for the rapid acquisition, information technology services for fast selling system, resulting in securities fund management institutions information system construction, basic is again after the production planning, a lack of unified planning, overall consideration. Also makes the whole industry system highlights two characteristics,

First, there are many kinds. For example, a PB business, many operating institutions have three sets of systems. A CRM system, many operating institutions basically have N sets…… ;

Second, each system is not big. In addition to the historical database for storing historical data, compared with the Internet and other industrial systems, the overall architecture is not complex, and the amount of data is also small.

In terms of external reasons, operating institutions need to connect with many market core institutions to carry out business, such as Shanghai and Shenzhen Stock Exchanges, China Clearing, Beijing, Shanghai, Guangzhou and Shenzhen branches of China Clearing, China Securities Finance, stock conversion, Shenzhen Stock Connect, Shanghai Stock Connect… .

The combination of internal and external causes makes operation and maintenance very complicated. For example, if you go down to the specific actions of each mouse click, it’s normal to take hundreds or thousands of steps every morning. Coupled with the industry’s zero-tolerance culture for system errors, the industry system operation and maintenance work can be said to be an extremely test of one’s patience, carefulness and concentration, and requires an extremely able to resist boredom.

Iv. Industry operation and maintenance experience

According to the author, the operation and maintenance mainly went through the following stages, but the industry operation and maintenance as a whole is still in the third stage.

The first stage is purely manual. In the early stage of industry development, or in the early stage of institutional exhibition industry, the system is less and depends entirely on manpower;

The second stage is the borrowing script. As systems proliferate, clever operations people start writing their own scripts to replace some of the manpower;

The third stage is automatic operation and maintenance. Later, more and more systems, more and more scripts, accidentally click the wrong script operation also happens, so someone will systematize these scripts, and enrich some functions, namely automatic operation and maintenance.

The combination of automated operations and real-time monitoring systems greatly saves manpower and reduces the risk of manual operation, demonstrating that fintech is the first productive force in the industry.

In the automated operation and maintenance market, the manufacturers and products represented at present include: ServiceJetiAuto platform of China Information, KC-AOM automated operation and maintenance system of Golden Securities, etc.

For large-volume business organizations, IT service processes are often standardized and constructed based on ITIL, the best international industry practice.

The increasingly automated and standardized operation and maintenance has brought a light to the dark world of operation and maintenance. At present, the vast majority of operating institutions (securities firms) have adopted automatic operation and maintenance.

The fourth stage, the DevOps stage. Before this, the development and operation and maintenance were completely isolated, which also led to the opposition and contradiction between each other. The development hoped to launch quickly, while the operation and maintenance hoped to find more thunder before launching. The contradictory KPI assessment content made the development and operation and maintenance quarrel a lot of things.

The concept of DevOps was introduced in order to unite development operation and maintenance and other related entities to jointly build the operation and maintenance system. It should be noted that industry development and operations are generally separate at the moment, and developers are generally not allowed to enter operations areas during a deal, except in exceptional circumstances.

The fifth stage is intelligent operation (Algorithmic operation, AIOps,Algorithmic IT Operations). In September 2016, Gartner defined AIOps for the first time, namely “an operation and maintenance method that automatically learns and summarizes rules from massive operation and maintenance data and even makes decisions through machine learning and other artificial intelligence algorithms”.

5. Feasibility analysis of intelligent operation and maintenance

Public information shows that currently in the securities industry, Galaxy Securities, CIC Securities in intelligent operation and maintenance have a certain attempt. The author believes that intelligent operation and maintenance of the industry has great potential, mainly for the following reasons:

One is the lack of failure analysis tools in the industry. Information security events occur in the industry, the time of fault disposal almost determines the final determination of the event, whether it is a general event, a major event or a major event.

The troubleshooting time window is very small, and time is precious. Intelligent O&M helps you quickly locate problems, find the root cause of problems, and provide decision-making suggestions.

Experienced o&M personnel know that there are often failures (or suspected failures) of unknown cause. Sometimes, restarting an application or server would be fine. However, when asked about the root cause, it is rarely clear that it is too expensive to explore the root cause or the problem is difficult to reproduce.

According to the author, more than 90% of the time spent on fault handling in the industry is spent on problem location and decision making. Intelligent operation and maintenance can greatly reduce these time and reduce the cost of information security incidents.

Second, the industry lacks the tools to predict failures in advance. The author’s opinion, a lot of details or information security event is attributed to working in hardware, resulting in a combination of hard and soft blind area, if we can based on all systems, middleware, database, disk, room temperature and humidity, and so on history log and fault cases of huge amounts of data, using machine learning algorithms, knowledge map, to predict failure ahead of time, will become very valuable.

Third, the industry operation and maintenance work needs to be more refined. The current industry is still developing, new application systems emerge in an endless flow, and the coupling degree between the systems is gradually increasing, the common phenomenon of small and scattered system will exist for a long time, automatic operation and maintenance instead of manual operation, monitoring system can timely reflect the normal state of the system; Over the course of a year, no information security incident is the biggest achievement of operation and maintenance personnel, but have you ever considered: your operation and maintenance for more than 200 days a year, the system status value is the same?

If no information security incident is defined as 60 points, is the overall state of the company’s system 90 points or 100 points every day? What about the state curve? What potential risk points are creeping up on us? This is something automation and monitoring cannot answer, while intelligent operations can.

Fourth, intelligent operation and maintenance is an excellent scenario for the application of fintech. Intelligence is based on data, data are generated by the system, the system is generated by the program, the program rules are fixed, fixed format, that is, master data, metadata fixed, facilitate the use of natural language processing technology (NLP) to process all kinds of log data.

Intelligent operation and maintenance relies on at least the following data: Log rooms of all kinds of environment such as temperature, humidity and so on, the system log and log (including operating system, database, middleware, etc.), all kinds of operation logs, etc., the cumulative amount of data is large, more important is putted forward, both case bank industry and the company over the years, can take advantage of the modeling analysis of machine learning, artificial intelligence, all kinds of algorithms.

Identify potential risks in advance; Accurately locate the root of the problem; The report is automatically generated after the event, and improvement suggestions are given to form a closed loop of operation and maintenance.

Fifth, the first embodiment of self-value of relevant posts. Current many operators have a large data, artificial intelligence and so on up to the technical post, if the department of eating, and very suitable for AI ground intelligence work is not to think, unwilling to make or maybe, and with what company executives believe you have the ability to do a good job, complete other areas of business intelligence?

Of course, intelligent operation and maintenance does face challenges. First, the investment of the company. Intelligent operation and maintenance is a cost-saving project on the surface, which requires more efforts to persuade the company’s leaders and gain support. Second, other developers need to cooperate with the problem. The acquisition and analysis of all kinds of software logs need other developers to open their permissions or provide support, which is also challenging.

Vi. Promotion of intelligent operation and maintenance

(I) Promote the principle of intelligent operation and maintenance

The principle to be followed for the promotion of intelligent operation and maintenance is to put safety first. The particularity of the securities fund industry determines that all kinds of operations based on the production system should be very careful, and all functions related to intelligent operation and maintenance should be tested and refined repeatedly.

Intelligent operation and maintenance is the ideal country of operation and maintenance. It is impossible to be in place in one step. Do not try to automate decisions at the beginning.

The last thing to remember is to follow blindly. Intelligent operation and maintenance is an action only after the operation and maintenance complexity of the operating organization is adjusted. The number and complexity of the system of most operating organizations are controllable, and the traditional automatic operation and maintenance can be done without killing chickens.

(2) The idea of promoting intelligent operation and maintenance

On the one hand, intelligent operation and maintenance is often suitable for business organizations with a relatively solid IT foundation. Of course, business organizations can also use the power of IT service providers. On the other hand, the market-neutral core organization can also provide more comprehensive intelligent operation and maintenance services (prediction, fault rapid location, decision suggestion, abnormal alarm aggregation, etc.) for the whole market operation organization by collecting various system logs and case libraries.

At present, intelligent O&M mainly solves problems such as fault risk prediction, alarm log aggregation, and fault disposal decision support. Most logs are processed in text type. It will use clustering tree, decision tree, random forest, support vector machine regression and other algorithms, as well as ARTIFICIAL intelligence technologies such as NLP to analyze text data and build knowledge base.

Vii. Author’s Suggestions (Essence)

According to the author’s experience, the following three points need to be paid special attention to in current industry operation and maintenance:

One is to remember reports during failures. Operation and maintenance in addition to daily operation, emergency, there is also a place that is easy to be ignored, that is, the reporting mechanism and process after the occurrence of information security time, which should be reported, what should not be reported, reported to whom, how to report and so on to figure out, it is best to do into the system, to avoid administrative supervision measures due to omission, concealment punishment;

Second, do not rely too much on automated operations. The automated operation and maintenance phase does not require any manual operations. When operation and maintenance adapt to and get used to automatic operation, once the system fails and does not operate manually for a long time, there will often be confusion of hands and feet, and even cause secondary failures.

In this regard, the author suggests that, in the holiday test or a specific period, selective manual operation, is also very necessary;

Third, attach importance to the foundation. Current industry overall information technology foundation is uneven, most are weak. Of course, good operation and maintenance is the foundation of the foundation.

Instead of bragging about smart XX every day, we should think more about how to consolidate the foundation of information technology, do a good job of information infrastructure, and clarify the boundaries of business compliance in the era of fintech.

The intelligent XX built on the basis of the flawed technology is not a big problem in the short term, but in the future operating institutions will pay even with interest (usury) or even higher price.

There are some debts we can pay late, but we can’t afford not to pay. Technical debt is one of them.

Note: This article is reproduced with authorization from the public account of The Way of Fintech, ID: BY_CSfCAolei