Microservice stability management

Stability governance is a process of fault management . From the perspective of fault management , It can be divided into fault prevention 、 Fault finding 、 Fault location 、 Fault stop loss and fault recovery 5 Stages , All works of stability construction are integrated into each stage of fault model . If stability failure is compared with fire , The stability working model is actually a fire – Set fire to – Fire extinguishing model .

The so-called fire prevention , Through various mechanisms and measures , Check out all possible hidden dangers in the system in advance , Prevent the occurrence of disaster ; Fire fighting refers to the actual occurrence of problems , Stop loss as much as possible , Reduce the impact of disasters , Resume the normal operation of the business as soon as possible ; Setting fire is similar to fire drill , Regularly simulate the occurrence of disasters , And formulate corresponding evacuation routes , Through the exercise, we can find out what still needs to be improved .

Microservices are being developed 、 test 、 Operation and maintenance 、 Capacity cost and other aspects have brought many problems , The use of container technology can greatly alleviate the problems caused by microservice architecture . The most revolutionary innovation of containers is the mirroring technology , It will apply 、 The basic library and environment are encapsulated together , As the cornerstone of microservice encapsulation and operation . Lightweight image technology as the delivery method of microservices .

• In a container environment , Because the image is light , After each change , You can quickly create the image corresponding to this change , At the same time, you can quickly create multiple identical test environments based on this secondary image , The container image encapsulates all the relevant details necessary to run the application , Such as application dependency and operating system . This makes it more flexible to migrate images from one environment to another , It strongly supports the testing under the rapid iteration scenario of microservices .

• There are many microservices under the microservice architecture , And every service changes very frequently , The workload of operation and maintenance is very large , With container mirroring , The environment can be delivered in advance . Pay more for each research and development 5% The amount of work , Exchange for O & M 200% The amount of work , It can accelerate the rapid deployment and landing of microservice changes .

The cloud architecture of microservices , There will be a big difference between the operation and maintenance level and the previous level , Infrastructure level , For example, deployment system 、 Configure the system 、 The monitoring system needs to be adapted to the cloud , There are also many unique problems in the process of cloud service .

• The container network is connected with the physical machine network

The cloud migration of microservices is a direct migration from physical machines to Kubernetes Container clusters . In order to reduce the Kubernetes Frequent change of calling mode in the process of containerization transformation , We follow an important principle in the migration process , Network interworking before and after migration , In this way, there is no need to change the calling method during the migration process , During the migration process, the business does not need to be aware .

• Identify differences in physical machine and container environments

Under the Cloud Architecture , Although there is no obvious difference between the use and before , But after all, they are two completely different environments . In the container environment, the corresponding character device is also configured under the host computer , However, it is not created separately for the container under the host , Therefore, during cloud migration , You should have a clear understanding of the differences between the container and the physical machine environment .

• Container resource isolation

In a container environment , Various service departments , When a container consumes more resources than a certain limit , If there is no perfect resource isolation mechanism , This will cause exceptions to other services on the same host . Refined resource isolation technology is a necessary prerequisite for the large-scale promotion of container technology .

• Disaster fault

After all, cloud service is a new environment , You may encounter various types of problems , Especially at the network level , Transition from hardware network to software defined network （SDN）, Although the flexibility is much stronger than before , but SDN The maturity and stability of the network are quite different from those of the previous hardware networks . You need to think in advance about the possible fault types , Set corresponding disaster recovery plan for each fault .