After using MQ message oriented middleware, I began to regret

V-xin：ruyuanhadeng get 600+ Page original boutique article summary PDF

One Previous review

Last article 《 Why use MQ Message middleware ? These questions must be taken down !》, I told you about the role of message oriented middleware in introducing system architecture , What problems are mainly solved .

The common practice scenario is ：

• Decoupling of complex systems
• Asynchronous call of complex link
• Peak clipping of instantaneous peak

Two The official start of the

This article tells you about , If you introduce message oriented middleware into the system architecture , What are the disadvantages ？

1 Reduced system availability

First, the overall availability of your system will definitely decrease , Let me give you an example , Let's take the previous picture to illustrate .

For example, in a core link , System A -> System B -> System C, And then the system C It's through MQ Call the system asynchronously D Of .

It's looks good , You use this MQ The asynchronous method solves the problem of poor performance of the core link .

But have you considered another question , It's the one you rely on MQ What if the middleware suddenly hangs ？ This is really not wishful thinking ,MQ、Redis、MySQL These components may hang .

Once your MQ Hang up , It leads to the interruption of the core business process of your system . Originally, if you didn't introduce MQ middleware , That is actually the call between some systems , But now you have introduced MQ, It will cause you to rely on one more . Once you have one more dependency , It will reduce your availability .

therefore , Once introduced MQ middleware , You have to think about this MQ How is it deployed , How to ensure high availability .

Even in complex high availability scenarios , You should also consider if MQ Once you hang up , Does your system have a backup technical solution , It can ensure the continuous operation of the system .

I wrote an article before , It's about MQ High availability guarantee scheme after hanging up .

If you are interested , You can refer to it ：

• 《 use RocketMQ Realize the final consistency scheme of reliable messages ,yyds！》

Through this article , Let's take a look at what we encounter in various scenes MQ High availability degradation scheme adopted for fault .

2 The stability of the system is reduced

The same picture above , Let's have a look again .

I wonder if you have found a problem , This link is in addition to MQ Besides the possible hidden danger of middleware failure , There may be other technical problems .

for instance , Inexplicably , System C Sent a message to MQ, As a result, the message was due to network failure and other problems , It's lost . This causes the system to D I didn't receive that message .

This is terrible , This will cause the system to D I didn't finish my task , At this time, the whole system may have business disorder , Data loss , serious bug, Poor user experience and other problems .

This is just one of them , In case the system C to MQ Send a message , Accidentally, I took a draught and repeatedly sent an identical , Cause the message to repeat , What to do at this time ？

May cause the system to D Insert a piece of data twice at once , Causing data errors , Generation of dirty data , Finally, it will lead to various problems .

Or if the system D Suddenly down for a few hours , Leads to inability to consume messages , As a result, a large number of messages are MQ There is a long backlog in middleware , What to do at this time ？

Even if the system D To restore the , We also need to process consumption data slowly .

So that's the introduction of MQ The second big problem of middleware , The stability of the system may decrease , Faults will increase , All kinds of messy problems can arise .

And once a problem arises , It will cause problems in the whole system . In order to solve all kinds of MQ Technical problems caused , Take a lot of technical solutions .

About this , We'll talk about it later with a special article MQ Solutions to these problems of middleware , Include ：

• Highly reliable message delivery （0 The loss of ）
• Message idempotent transmission （ Never repeat ）
• Online fault handling of millions of messages backlog

3 Distributed consistency issues

Introduce message middleware , There is also the problem of distributed consistency .

for instance , For example, the system C Now you have successfully processed your local database , Then sent a message to MQ, System D It is indeed consumption .

But unfortunately , System D Failed to operate your local database , What should I do at this time ？

System C succeed , System D failed , It will cause the overall data of the system to be inconsistent .

So at this time, we need to use the distributed transaction scheme of reliable message final consistency to ensure .

About this , You can refer to a previous article ：

《 use RocketMQ Realize the final consistency scheme of reliable messages ,yyds！》

We have described in detail the scenario of asynchronous calls between systems , How to adopt distributed transaction scheme to ensure its data consistency .

3、 ... and summary

Last , Let's make a simple summary .

Answer this question well in the interview , First of all, be familiar with MQ The advantages and disadvantages of this technology . Understand what problems he was introduced into the system to solve , But what problems will he bring .

Besides , For the introduction of MQ in the future , Is there any plan design for the problems that may be caused by himself , To ensure high availability of your system 、 Highly reliable operation , Ensure data consistency . This also makes corresponding preparations .

V-xin：ruyuanhadeng get 600+ Page original boutique article summary PDF