Some cold thoughts behind the popularity of microservices, middle office, RPA and low code

Last year I went to 2021 South China CIO The conference , It is found that there is basically a hot spot at the conference every year , For example, this year's hot spot is the low code development platform .

We can review the hot spot changes in recent years .

• 17-18 year ： Microservices
• 18-19 year ： Zhongtai
• 19-20 year ：RPA, Digital marketing
• 20-21 year ： Low code , Cloud native

From my own article output , Basically, it is also in line with this general trend . For example, I 17 More articles on the design and implementation of microservice architecture will be written around ,18,19 Many articles on the construction of China Taiwan were output in . In recent years, the focus has been on cloud native overall solutions and low code development platforms .

stay IT industry , Various emerging technologies emerge one after another , All kinds of word making abilities of Internet manufacturers are also quite strong , It also enables new hot spots and new technologies to emerge every year .

similar 19 Low code development may have just sprung up in , This year, we can see all kinds of manufacturers making low code platforms , Whether it's doing SaaS Service , Or tradition toB Implemented , Or around Tencent , Nail ecological . It is estimated that there are at least 100 Enterprises are making low code products . As a result, the whole industry with low code is also in a situation of competition and chaos .

In the first half of the year ,ThoughtsWork Xu Hao wrote an article , When it comes to low code platforms, they are the cancer of the industry , First of all, I personally disagree with this view , Any new thing has its value and reason of existence , It's impossible to solve all your problems . Instead, we should adopt the most appropriate method at the right time .

The low code platform itself is a good thing , But many vendors boast that low code platforms are omnipotent , No matter how complex the system and rules are, they can have zero code , Drag and drop to achieve , This is not talking about martial virtue at all . So low code platform is not an industry cancer , On the contrary, it is the nonsense and boasting in the competition of manufacturers that are the cancer of the industry .

Looking back, the concept of middle stage is the same .

Zhongtai itself is a good concept and thought , Emphasize the sinking of common business capabilities of enterprises , Then a reusable business capability center is formed and provided to the upper application , So that the upper application can be developed flexibly and agile .

There is nothing wrong with this idea itself .

But a lot of manufacturers in China and Taiwan , Especially many Internet entrepreneurs , Blindly exaggerating the role of Zhongtai , Draw big cakes for enterprises , If you have a middle stage, you can do everything , What the enterprise already has IT The system should also be changed all over again , To build a large and comprehensive middle office capacity, rather than considering how to retain the legacy of the enterprise IT assets . These also led to the final construction failure of a large number of China Taiwan projects , Or it didn't work as expected .

It's not that Zhongtai's thinking is bad , But the manufacturer exaggerates the propaganda and fails to achieve the final effect and goal , As a result, users continue to bite back , This can't blame the user, but the manufacturer .

The same is true when we come back to micro services .

Go backwards 3 To 5 year , It is estimated that many enterprises have also been killed by microservices .

The reason is also simple , A single application itself runs well , Finally split into 20 Multiple microservices , This leads to complex integration among multiple microservices , Distributed transactions are out of control , Subsequent troubleshooting is difficult , A series of problems such as difficulties in operation and maintenance monitoring .

This in itself is not the wrong idea of microservice , It's not the application, it's the wrong application .

One is that the enterprise has not reached a certain IT Blindly use micro services when managing management and control capabilities , The second is the unreasonable splitting of microservices in the early architecture modeling stage, resulting in too detailed splitting , Or close coupling between separated microservices .

The idea of micro service itself should not carry this pot .

Last year I participated in South China CIO At the conference ,RPA The robot fire is a mess , It's said that there are some problems this year RPA The manufacturer or team has been disbanded .

that RPA Whether machine automation is good or not ？

Same thing , The existence of any new thing has its reason .RPA The integration of robot and automation technology solves the problem of difficult bottom integration of traditional business system , Automate repetitive work .

There is no problem with this idea . It must have its application scenario and application value . But realize that RPA More of a compromise , Not the target solution .

Why do you say so ？

If a party a enterprise itself has the ability to do the data integration and interface integration between the underlying business systems , But you're lazy and don't do it yourself , But through the upper layer RPA To solve the problem . So it's an obvious way to treat the symptoms rather than the root cause .

A big tree , Multiple tree roots at the bottom of itself should be integrated and staggered to form a joint force , The branches supporting the upper layer are luxuriant . But now the integration between the roots of the underlying tree is not done , Pull the rope on the branches and leaves in front , Bundle line . Although this can solve the problem temporarily , But in the end, the more difficult it is to develop and grow behind this tree , It's not impossible to fall suddenly one day .

So when I rethink these hot concepts , Some key thoughts are given and summarized as follows .

Microservices

Reaffirm the principle , That is, you should not arbitrarily split the micro service without clear requirements . Even if you use a microservice development framework , You can also not do a big split . For most enterprises , The actual amount of business concurrency has not reached the point where microservicing is necessary to solve the problem .

secondly , The application expansion gives priority to the expansion method under the traditional monomer mode , Similar to cluster expansion , Database read and write separation, etc . You can also expand by sub organization level .

Zhongtai

If your enterprise itself has a certain foundation for information construction , So the best way to build a middle platform is to build on the existing legacy IT Sort out the reusable business capabilities in the system , be based on SOA To build a business service sharing center . Instead of building a new middle platform .

For the data center , If there is no fine-grained micro service splitting , The problem of data feeding business does not need to be solved by the data center , It can be solved directly in the business center or in the traditional legacy business system . Therefore, traditional enterprises build data platforms , Not to pursue the openness of data services and feed back the business , But the analysis and utilization after data integration , The idea may still be traditional BI System construction ideas .

RPA Robotics and automation

about RPA It's a compromise, not a target . When enterprises are faced with many scenes where it is difficult to integrate the underlying interfaces of legacy systems , May adopt RPA To solve the problem of automatic collaboration of repetitive work . But if there are conditions , It still focuses on the integration of bottom data and interfaces rather than the collaborative integration of upper interfaces .

RPA Don't get bigger and bigger , This later maintenance will be a big problem . One is that the core logic itself is not clear , One is that the underlying business system itself is also in an unstable state of change .

Low code development platform

In the industry standard specification of low code platform itself , Maturity is not reached before . Enterprises should not put the core business system on the low code development platform .

Low code platform enterprises can make some attempts , Can be similar to OA, project management , Operation and maintenance management and other partial work orders and process business systems are built on a low code development platform , Accumulate relevant practical and application experience .

Finally, when choosing a low code development platform, we should consider not to be kidnapped by the platform manufacturer , One of the basic requirements of any application developed by a low code development platform is that it can run without a low code platform , And have sufficient high availability and scalability requirements . Finally, I would like to share with you some digital work done in the early years The new retail saas Commercial products , Now it's all open source , You can study with small partners gitee On the search weiit-saas. weiit-saas Is a Java Open source project , Belong to weiit The team develops its own products , Intended to be packaged through technology , Let the enterprise do not need code development , Help enterprises generate small programs with one click 、 official account , Let the enterprise have the independent brand self operated Mall . Product competitors 《 I like it 》、《 Micro Alliance 》.