Shulou(Shulou.com)06/02 Report--

This article will explain in detail what is Hystrix, the content of the article is of high quality, so the editor will share it with you for reference. I hope you will have a certain understanding of the relevant knowledge after reading this article.

Hystrix coordinates the framework of high availability of various services in distributed systems, and is an important member of SpringCloud architecture. Hystrix mainly solves the problems of service avalanche, service monitoring and so on.

Hystrix function introduces the function of resource isolation and current restriction.

The main purpose is to prevent the collapse of the dependent service from affecting the consumption of its own service resources, and to use isolation to limit the maximum thread resources that can be used to invoke the dependent service. If a service crashes and does not control resource isolation, it will trigger a service avalanche and many upstream services will collapse collectively.

In addition, in the case of large concurrency, it can also reduce the traffic of dependent service requests and avoid excessive load of dependent services.

Principle

Hystrix encapsulates calls to external services and a series of processing into a Commond object, which is divided into HystrixCommand (direct invocation command) and HystrixObservableCommand (observer subscription command), which can register callback events. Both can be called asynchronously or synchronously (Future is returned asynchronously, and get blocking can be called later to get the result).

Commond objects isolate resources in two ways, one is to use thread pools (commonly used), and the second is to use semaphores. Both of them limit the number of concurrent threads called by Commond. The difference is that semaphores are only a thread that allows concurrent calls to command and cannot record timeout information and so on.

Configuration item

Select Commond resource isolation strategy, thread pool (commonly used) or semaphore

Specify the command name, group. The name defaults to the class name (a command usually corresponds to a service-dependent interface call), and all command under group share a thread pool or are isolated by a semaphore by default (group usually corresponds to a dependent service).

Specify the command thread pool, and the key for HystrixThreadPool defaults to the group name, or you can specify the HystrixThreadPool to use manually.

Specifies the number of core threads in the thread pool, which defaults to 10.

Specifies the thread pool queue length, which defaults to-1 unbounded queue. If the value is exceeded, the request is rejected.

Specifies the thread pool reject threshold, which defaults to 5, and rejects the request if it exceeds it.

The maximum number of concurrent visits can be set under semaphore isolation. The default is 10.

Principle of requestCache- request caching

In a web call main thread, the cache is established for the command of the main thread calling the service, so that if the call depends on the same parameters of the service, the result can be obtained directly from memory. Since I think it is not very useful, I will not introduce it much, mainly initializing the HytrixRequestContext in the interceptor (this cache data should be visually in the thread local variable), and then implementing the getCacheKey method in Command to return the key of cache. When the key is the same, directly try to get the result of the same key returned after the previous call.

Principle of fallback- service degradation

The getFallBack method (HystrixCommand) or the resumeWithFallback method (HystrixObservableCommand) is implemented through Command, and the degraded logic is implemented, and the degraded result is returned locally.

Conditions for demotion

Thread pool or semaphore reject

Call timeout

Execute error report

The circuit breaker is turned on

The usual logic of demotion

Fetch older data from the local cache

Returns the default value

Note: downgrade logic can set parameters to limit the number of concurrent threads (based on semaphores) circuit breaker- circuit breaker principle open condition

The call timeout reaches a certain percentage.

The execution of error reports reaches a certain proportion.

Thread pool or semaphore reject to a certain proportion

Half-open state

After a certain period of time, it will switch to the half-open state and let 1 request pass to see if it can be returned normally.

Open state

If the request allowed to pass in the half-open state is successful, turn off the circuit breaker

Related configurable item

Whether to turn on the circuit breaker (on by default)

Minimum number of requests in the circuit breaker time rolling window (default 20)

How many anomalies (including timeouts, errors, rejections) (default 50%) turn on the circuit breaker

How long does it take to enter the half-open state (default is 5000 milliseconds)

Whether to force the circuit breaker to open

Whether to forcibly close the circuit breaker

Timeout configuration (default 1000 milliseconds)

Whether to turn on the timeout mechanism (on by default)

On what is Hystrix to share here, I hope that the above content can be of some help to you, can learn more knowledge. If you think the article is good, you can share it for more people to see.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.