Cartoon: what is distributed transaction?

————— the second day —————

————————————

If there is no distributed transaction

In a series of microservice systems , If there are no distributed transactions , What will happen ？ Let's take the trading business commonly used in the Internet as an example ：

The figure above contains two independent microservices, inventory and order , Each microservice maintains its own database . In the business logic of the trading system , A product needs to call inventory service before placing an order , Deduct inventory , Call the order service , Create order records .

Under normal circumstances , The two databases were updated successfully , The data on both sides is consistent .

however , Under abnormal circumstances , It's possible that the inventory deduction has been completed , The subsequent order record failed to be inserted for some reason . This is the time , The data on both sides will lose their due consistency .

What is distributed transaction ？

Distributed transactions are used to ensure data consistency between different nodes in a distributed system . There are many ways to implement distributed transactions , The most representative is by Oracle Tuxedo The system proposes XA Distributed transaction protocol .

XA The agreement contains Two-phase commit （2PC） and Three stage commit （3PC） Two implementations , Here we focus on the specific process of two-stage submission .

In world of Warcraft , Copy group play BOSS When , In order to facilitate the cooperation between the team leader and the team members , The captain can initiate a “ Position confirmation ” The operation of ：

When the team member receives the confirmation of position , If it's in place , Just choose “ yes ”, If it's not in place yet , Just choose “ no ”.

When the captain receives confirmation of everyone's position , It's going to be released to all the players , Tell them to start fighting BOSS.

Corresponding , When the captain initiates the confirmation of position , It's possible that some of the players are not in place yet ：

That's the world of Warcraft BOSS The confirmation process of . This process and XA The two-phase commit of the distributed transaction protocol is very similar .

that XA What exactly does the agreement look like ？ stay XA There are two roles in the agreement ： A business coordinator and Transaction participants . Let's take a look at the interaction process between them ：

The first stage ：

stay XA The first phase of distributed transactions , The node that acts as the transaction coordinator will first send Prepare request .

After receiving Prepare After the request , Each participant node will update the data related to the transaction , write in Undo Log and Redo Log. If the participant performs successfully , Do not commit the transaction for the time being , Instead, it returns... To the transaction coordination node “ complete ” news .

When the transaction coordinator receives a return message from all participants , The whole distributed transaction will enter the second phase .

The second stage ：

stay XA The second phase of distributed transactions , If the transaction coordination node has received all the data before, it will return in a positive direction , Then it will issue... To all transaction participants Commit request .

Receive Commit After the request , The transaction participant nodes will commit their own local transactions , And release the lock resource . When the local transaction is committed , Will return to the transaction coordinator “ complete ” news .

When the transaction coordinator receives the “ complete ” feedback , The whole distributed transaction is completed .

What is described above is XA The forward process of two-phase submission , Next, let's take a look at the process of handling the failure situation ：

The first stage ：

The second stage ：

stay XA The first stage of , If a transaction participant feeds back a failure message , Indicates that the local transaction execution of this node is not successful , Must roll back .

So in the second stage , The transaction coordination node sends Abort request . Received Abort After the request , Each transaction participant node needs to roll back the transaction locally , The rollback operation follows Undo Log To carry out .

That's all XA The detailed process of submitting the agreement in two stages .

XA The shortcomings of two-phase submission

XA What are the shortcomings of the two-stage submission ？

1. Performance issues

XA The protocol follows strong consistency . During transaction execution , Each node occupies the database resources , Only when all nodes are ready , The transaction coordinator will notify the commit , Release resources after participants submit . This process has very obvious performance problems .

2. Coordinator single point of failure problem

The transaction coordinator is the whole XA The core of the model , Once the transaction coordinator node fails , Participants cannot receive commit or rollback notification , Participants will always be in an intermediate state, unable to complete the transaction .

3. Inconsistencies caused by missing messages .

stay XA The second phase of the agreement , If there is a local network problem , Some transaction participants received a commit message , Another part of the transaction participants did not receive the commit message , This leads to data inconsistency between nodes .

If you avoid XA All kinds of problems in the two-stage submission ？ There are many other distributed transaction schemes to choose from ：

1.XA Three stage commit

XA The three-phase commit adds... To the two-phase commit CanCommit Stage , And introduced the timeout mechanism . Once the participants don't get a call from the coordinator commit request , Will be automatically local commit. This effectively solves the problem of coordinator single point of failure . But the problems of performance and inconsistency are still not fundamentally solved .

2.MQ Business

Using message oriented middleware to update the second half of the transaction asynchronously , Achieve the ultimate consistency of the system . This approach avoids things like XA Protocol performance issues .

3.TCC Business

TCC The business is Try、Commit、Cancel Abbreviations for the three instructions , Its logical pattern is similar to XA Two-phase commit , But the way to implement it is to implement it artificially at the code level .

Some supplements ：

. This cartoon is purely entertainment , Please cherish your current work as much as possible , Don't imitate Xiao Hui's behavior .

—————END—————