Service Dependency Graph

Store and display microservice dependency graph.

Projects Description

Service Stats

It's responsible for receive microservices data and returns that data in a graph format to be displayed by the Service Viewer project. All data is stored using the current time as an identification, so this way is possible to view data trough time.

This data in JSON format is sent through a HTTP POST to /calling URI, and it's represented by:

  • Caller Name (Microservice that performs the call)
  • Called Name (Microservice that receives the call)
  • Called Endpoint Name (Microservice's endpoint that receives the call)



Important Note: this process return 200 (OK) instantly (it's basically a request and forget method), this way the clients don't need to wait for the whole process to be executed.


$ sbt test


It requires a Redis instance running locally.

$ sbt run

Insert Dummy Data

It requires curl command to run this script.

Please, do not forget to run the service-stats through SBT before executing this script.

Just run the script:

$ ./

It will perform some HTTP post requests to http://localhost:8080, after that you can see the graph using the service-viewer project.

Technology Used

  • Scala language
  • Akka HTTP
  • Scalatest
  • Redis (rediscala)

How it works internally

Storing data

All data is stored into Redis using HASH format to represent graph data.

The key for Node/Vertex is (HH is the hour and mm is the minute). Example: nodes:17.41

The key for Link/Edge is (HH is the hour and mm is the minute). Example: edges:17.41

Each edge contains a value that represents how many requests it received at a specific time, and it can be visualized as a label rep/min by the service-viewer graph).

The node's index is generated by a Redis String using the inc command.

Displaying Data

Currently is only possible to display a graph for a specific time (hour:minute).

What is missing (Things to do)

As it does not intend to be a time series graph, it would be good to determine how many time all data that represents a specific time will be available. Adding TTL to those data could fix that issue, so this way there'll be no conflict among dates.

Service Viewer

It's responsible for displaying the graph in a browser.

It's just an HTML file (service-viewer/index.html) that you can open directly in your browser.

This HTML uses: Jquery and D3.js


how it looks like

Built With

Share this project: