opentracing-infrastructure-graph

Visualizing infrastructure topology via OpenTracing instrumentation.

This application uses the following libraries to extract the topology metrics to Prometheus via OpenTracing:

• opentracing-metrics-tracer
• opentracing-auto

Requirements

• Docker

Run Prometheus

Modify: /prometheus-data/prometheus.yml, replace 192.168.0.10 with your own host machine's IP.
Host machine IP address: ifconfig | grep 'inet 192'| awk '{ print $2}'

docker run -p 9090:9090 -v "$(pwd)/prometheus-data":/prometheus-data prom/prometheus -config.file=/prometheus-data/prometheus.yml

Open Prometheus: http://http://localhost:9090

Getting started

It will start three web servers and simulate a service call chain:
server1 calls server2 and server3 parallel.

npm start
curl http://localhost:3001

Metrics between services

parent_service="unknown" label means that the request initiator is not instrumented (Prometheus scraper, curl, etc).

Throughput

Prometheus query:

sum(rate(operation_duration_seconds_count{name="http_server"}[1m])) by (service, parent_service)  * 60

95th response time

Prometheus query:

histogram_quantile(0.95, sum(rate(operation_duration_seconds_bucket{name="http_server"}[1m])) by (le, service, parent_service)) * 1000

Infrastructure topology

Data comes from Prometheus.
Uses vizceral.

npm run start-client
open http://localhost:8080

Future

• add databases
• show latency