Distributed link tracking skywalking Foundation

Introduce

SkyWalking Born for super large scale . Whether your microservice is in the service grid （Service Mesh） Under the architecture , It can provide high-performance and consistent monitoring .

Application performance monitoring tools for distributed systems , Dedicated to micro Services 、 Cloud native architecture and container based (Docker. K8s、Mesos) Architecture and Design . Provide distributed tracking 、 Service grid telemetry analysis 、 Measurement aggregation and visualization integration solution .

The background
In microservice system , As the business grows , The system will get bigger and bigger , Then the call relationship between services becomes more and more complex . One HTTP The request will call several different microservices to process the final result , During this call , The request may fail due to the high network delay or sending error of a service , This is the time , The monitoring of request calls is particularly important .SpringCloud Sleuth It provides the solution of distributed service link monitoring .

The production environment call interface reports an error , How to make positioning problems ?
1、elk+kafka, It just reduces the cost of developers searching logs .
2、 If the production interface call chain reports an error : Need to communicate between teams
3、 Distributed service tracking system —— Solve the call chain error in the production environment , Reduce communication costs between teams .
Distributed service tracking system ： The bottom layer contains error logs ( Abnormal interface ) 了

Commonly used distributed service tracking system
1、sleuth+Zipkin;
2、 Alibaba's eagle eye system ;
3、skywalking

Be careful ： If the call chain is too long , It may cause the problem of calling timeout .

The principle of distributed service call chain tracing

every time RPC Remote call requests will generate a spanid Record every time rpc Requested information , One more traceid Globally unique id;

Tranceld： overall situation id, Record the relationship of the entire request chain
Spanid： unit id, Record - Secondary call relationship
Parentld： Paternal id

SkyWalking

SkyWalking Is an open source observation platform , Used to collect data from infrastructure such as services and cloud native 、 analysis 、 Aggregate and visualize data ,SkyWalking It provides a simple way to clearly observe distributed systems , You can even observe systems across different clouds ,SkyWalking More like a modern application performance monitoring (Application Performance Monitoring, namely APM) Tools , For cloud native , Designed based on containers and distributed systems .

Related name concepts

1. probe ： Probes may be different based on different sources - What kind of , But it's all about collecting data , Format data as SkyWalking Applicable format . For example, in Java In this case, we do bytecode implantation , No intrusive collection , And pass HTTP perhaps gRPC Send data to the back end of the platform through Agent technology Proxy main function .
2. Platform backend ： It is a backstage supporting cluster mode , For data aggregation 、 Data analysis and the process of driving data flow from probes to user interface . The back end of the platform also provides various pluggable capabilities , Such as data from different sources ( Tathagata comes from Zipkin) format , Different cluster and storage system management . You can even use the observation analysis language for custom aggregation analysis .
3. Storage : It's open , You can choose an existing storage system , Such as ElasticSearch、 H2 or MySQL colony ( Sharding-Sphere management ), You can also choose to implement a storage system yourself .
4. The user interface ： That is to say SkyWalking The visual interface of ,UI Very cool and powerful , It can also be customized to match your existing backend .
5. SkyWalking It provides solutions for observing and monitoring distributed systems in many different scenarios . Such as Java、C# And Node. js Provide automatic language probe , No intrusive collection . Also for some compiled languages C++、G0 Etc. provides manual management SDK( Not yet supported ). besides , You can also use service grid based probes to collect data , To help understand the entire distributed system .
6. service (Service) ： Represents a series or set of workloads that provide the same behavior for requests . In the use of a dotting agent or SDK When , You can define the name of the service . If not defined ,SkyWalking Will use the name you defined on the platform , Such as Istio
7. Service instance (Service Instance)： Each of the above set of workloads is called an instance . It's like Kubernetes Medium pods equally , An instance of an operating system is not necessarily a service . But when you're using a dotting agent , A service instance is actually a real process on the operating system .
8. Endpoint (Endpoint) ： For the request path received by a specific service , Such as HTTP Of URI Path and gRPC The class name of the service + Method signature
9. Sum up ,SkyWalking The advantages are as follows ： Multiple monitoring means , Language probes and service grids (Service Mesh)、 modularization ,UI、 Storage 、 There are many mechanisms for cluster management 、 Support alarm 、 Excellent visualization scheme .

Four characters

1. skywalking agent Associated with the business system - rise , Responsible for collecting all kinds of monitoring data ;
2. oapservice It's responsible for processing monitoring data , Like accepting skywalkingagent Monitoring data , And stored in the database ( for example elasticsearch、mysql secondary ) ; Accept skywalking webapp Front end request for , Query data from a database , And return the data to the front end . Skywalking oapservice Usually in the form of clusters ;
3. skywalking webapp, Front end interface , Used to show data ;
4. A database used to store monitoring data , such as mysql、elasticsearch etc. ;

SkyWalking Environmental installation

Reference documents ：https://skywalking.apache.org/zh/2020-04-19-skywalking-quick-start/#

1、 download apache-skywalking-apm-8.9.1.tar.gz Installation package
After decompression , File directory

2、 Enter into bin Catalog ( Double click directly to start startup.bat )
startup.bat The order is as follows ： It will start SkyWalking OAP Service and SkyWalking UI service

perform startup.bat command

3、 start-up oapService And our webappService
4、 View the management interface http://127.0.0.1:8080/
My computer has done local domain name mapping , So the domain name is different .