Getting started with Zipkin

One 、 Introduce

Zipkin It's an open source Distributed real-time data tracking system （Distributed Tracking System）, be based on Google Dapper From the design of the paper , from Twitter Company development contribution . Its main functions are Gather Real time monitoring data from various heterogeneous systems .

Two 、 framework

Zipkin In two parts ：

• zipkin server
• zipkin client
  

zipkin server

Used as data acquisition and storage 、 Data analysis and display .

It mainly includes four modules ：Collector、Storage、API、UI

• Collector: Collector as , It is mainly used to deal with An external system Sent by Tracking information , Put this information transformation by Zipkin Internally processed Span Format , To support subsequent storage , analysis , Display and other functions
• Storage: Store received or collected data ( Support Memory,MySQL,Cassandra,ElasticSearch…), Default stored in Memory in
• API： Provide external query interface
• UI： Provide external Web Interface , adopt UI Component users can easily and intuitively query and analyze tracking information

zipkin client

Completed the generation and reporting of tracking data

3、 ... and 、 Basic concepts

Annotation

Record events that occur at a certain point in time , There are four types of events :

• cs(Client Send): The client sends request
• sr(Server Recevied): Server receive Request to start processing
• ss(Server Send): Server side Finished processing , send out To client
• cr(Cilent Recevied): Client reception Response to the server , End of request

BinaryAnnotation

Provide some additional information （ such as ：sessionID、userID、userIP、 Abnormal etc. ）, The required data can be Key value pair In this field

Span

Trace the basic structure of service invocation , By a group Annotation and BinaryAnnotation form , The processing trace for each service is a Span.
When a request is received and processed , Just create a Span, If the current request is sent from the upstream , The new Span The parent node of is set to the one requested in the previous step Span.

Trace

By a group traceId identical Of Span Connect in series to form a Trees structure , Indicates a call link .

Requests arrive at the... Of the distributed system entrance At the end , Generate a unique identifier traceId, During the internal circulation of the distributed system , The framework always keeps passing that unique value

Four 、 Information structure passed between services

Trace The basic information of needs to be transferred between upstream and downstream services , The following information is required ：

• Trace ID： start ( root ) Service generated TraceID
• Span ID： Generated when calling downstream services Span ID
• Parent Span ID： Father Span ID
• Is Sampled： Whether sampling is required
• Flags： Tell downstream Services , Whether it is debug Reqeust

5、 ... and 、 Service tracking process

1. Application initiate request , after Trace Frame interception
2. Of the current call chain Trace Information added to HTTP Header in
3. Record the current timestamp
4. take Trace dependent header Carry the information , Send a request
5. End of call , Record the current time spent
6. The information generated in the above steps is collected into one span, hold span Upload to zipkin Of Collector modular

 
┌─────────────┐ ┌───────────────────────┐  ┌─────────────┐  ┌──────────────────┐
│ User Code   │ │ Trace Instrumentation │  │ Http Client │  │ Zipkin Collector │
└─────────────┘ └───────────────────────┘  └─────────────┘  └──────────────────┘
       │                 │                         │                 │
           ┌─────────┐
       │ ──┤GET /foo ├─ │ ────┐                   │                 │
           └─────────┘         │ record tags
       │                 │ ───┘                   │                 │
                           ────┐
       │                 │     │ add trace headers │                 │
                           ───┘
       │                 │ ────┐                   │                 │
                               │ record timestamp
       │                 │ ───┘                   │                 │
                             ┌─────────────────┐
       │                 │ ──┤GET /foo         ├─ │                 │
                             │X-B3-TraceId: aa │     ────┐
       │                 │   │X-B3-SpanId: 6b  │   │     │           │
                             └─────────────────┘         │ invoke
       │                 │                         │     │ request   │
                                                         │
       │                 │                         │     │           │
                                 ┌────────┐          ───┘
       │                 │ ─────┤200 OK  ├─────── │                 │
                           ────┐ └────────┘
       │                 │     │ record duration   │                 │
            ┌────────┐     ───┘
       │ ──┤200 OK  ├── │                         │                 │
            └────────┘       ┌────────────────────────────────┐
       │                 │ ──┤ asynchronously report span     ├──── │
                             │                                │
                             │{                               │
                             │  "traceId": "aa",              │
                             │  "id": "6b",                   │
                             │  "name": "get",                │
                             │  "timestamp": 1483945573944000,│
                             │  "duration": 386000,           │
                             │  "annotations": [              │
                             │--snip--                        │
                             └────────────────────────────────┘

6、 ... and 、 install Zipkin Server side

Two way ：

1. Build yourself Zipkin Server side
2. Use the official compiled jar package

The first method , I read a lot of articles on the Internet , Only written. Zipkin The dependency of does not provide a version number , As a result, dependencies always fail ; The version number is written because of and SpringBoot The service cannot be started due to incompatible versions ;

Finally, I know , stay Springboot 2.0 in the future , Officials no longer recommend building their own Zipkin modular , It is recommended to use the official compiled Jar package

2.1 download jar package

zipkin-server-2.12.9-exec

2.2 Start the service

open cmd, In the corresponding jar Enter the command in the directory where it is stored ：

java -jar  The full name of the document 

Access port 9441, The following interface is displayed, which means the installation is successful