docs: update readme for nginx example

nginx-log-metrics/README.md

@@ -1,72 +1,100 @@
-# Nginx log metrics demo
+# The Nginx Observability Demo for GreptimeDB
 
-## Quick Start
+This is a demo for [GreptimeDB](https://github.com/greptimeteam/greptimedb) that
+takes Nginx Observability as an example. We have following features covered:
 
-Build and start with proxy:
+- Using GreptimeDB as **a scalable Prometheus backend** for Nginx metrics
+- Using GreptimeDB to **parse and store structured events** from Nginx access
+  logs and python server logs
+- Generating metrics from Nginx access log using GreptimeDB's **continuous
+  aggregation**
+- Running **a single joined query** from both metrics and events
+- Visualizing everything above from Grafana
+  - Tree map to show the user-agent corresponding platform that is extracted
+    in pipeline
+  - Log details of newest 50 nginx access log
+  - Two time-series panels shows flow continuous aggregation and conditional
+    aggregation
+  - Table shows top 10 slow trace IDs
+  - Joined log details shows top 10 slow logs based on provided search strings
+    to trace ID. Both Nginx and server sides details are included.
+
+![screenshot](https://raw.githubusercontent.com/GreptimeTeam/demo-scene/blob/main/nginx-log-metrics/screenshot.png)
+
+## How to run this demo
+
+Make sure you have `git`, `docker` and `docker-compose` installed. To run this
+demo:
 
 ```shell
-docker compose build --build-arg https_proxy=http://172.17.0.1:7890
-docker compose up -d --force-recreate
+git clone [email protected]:GreptimeTeam/demo-scene.git
+cd demo-scene/nginx-log-metrics
+docker compose up
 ```
 
-## Structure
+It can take a while for the first run to pull down images and also build
+necessary components.
+
+Once it's up and running , open your browser at `http://localhost:3000` and
+login with username/password `admin`/`admin`. Check the pre-built dashboards.
+
+You can also access GreptimeDB if you have `mysql` installed. Just run `mysql -h
+127.0.0.1 -P 4002` to connect to the database and using SQL query like `SHOW
+TABLES` as a start.
+
+## How it works
+
+This demo is to simulate a typical setup of a python based web application.
+
+The data plane includes:
+
+- A python web application
+- Nginx
+- A web client generates random traffic
 
-<!-- This chart is generated by https://derlin.github.io/docker-compose-viz-mermaid/ -->
+The observability plane includes:
+
+- GreptimeDB, of course, as the storage of all data
+- Vector, Prometheus and its exporter for moving logs and metrics to
+  GreptimeDB
+- Grafana for dashboarding
+- Some one-shot containers for initialization
+
+The topology is illustrated in this diagram. One-shot containers are ignored.
 
 ```mermaid
 flowchart TB
-  Vgrafanaprovisioning{{./grafana_provisioning}} x-. /etc/grafana/provisioning .-x grafana
-  Vconfigdata{{./config_data}} x-. /config_data .-x initdatabase[init_database]
-  Vconfigdata x-. /config_data .-x initpipeline[init_pipeline]
-  Vconfigdata x-. /config_data .-x vector
-  Vlogs([logs]) x-. /logs .-x vector
-  Vlogs x-. /var/log/nginx .-x nginx
-  grafana --> greptimedb
-  initdatabase --> greptimedb
-  initpipeline --> initdatabase
+  pyclient --> nginx
   nginx --> server
-  nginxexporter[nginx_exporter] --> greptimedb
+  grafana --> greptimedb
   nginxexporter --> nginx
-  pyclient[py-client] --> nginx
-  vector --> initpipeline
+  prometheus --> nginxexporter
+  prometheus --> greptimedb
   vector --> nginx
-  P0((4000)) -.-> greptimedb
-  P1((4001)) -.-> greptimedb
-  P2((4002)) -.-> greptimedb
-  P3((4003)) -.-> greptimedb
-  P4((3000)) -.-> grafana
-  P5((9113)) -.-> nginxexporter
-
-  classDef volumes fill:#fdfae4,stroke:#867a22
-  class Vgrafanaprovisioning,Vconfigdata,Vconfigdata,Vconfigdata,Vlogs,Vnginxconfnginxconf,Vlogs,Vconfigdataprometheusgreptimedbyml volumes
-  classDef ports fill:#f8f8f8,stroke:#ccc
-  class P0,P1,P2,P3,P4,P5 ports
+  vector --> server
+  vector --> greptimedb
 ```
 
-## Components
-
-This demo will start a nginx in front of a python server, and use a client to request the nginx.
-
-In this system, we have configured the following observability components:
-- nginx-export to collect and export basic metrics from nginx
-- vector to collect and push nginx access log
-- prometheus to scrape metrics from python server and nginx
-- python server send its logs direct to GreptimeDB
-
 All of those logs and metrics are stored in GreptimeDB, with:
-- a pipeline to fashioning the nginx access log. The config file is at [greptime_pipeline.yaml](./config_data/greptime_pipeline.yaml)
-  - string decomposition, field parsing or renaming etc can be defined in pipeline process the semi-structured log
+
+- A pipeline to fashioning the Nginx access log. The config file is at
+  [greptime_pipeline.yaml](./config_data/greptime_pipeline.yaml)
+  - string decomposition, field parsing or renaming etc can be defined in
+    pipeline process the semi-structured log
   - regex match and remapping
-- a flow to continuous aggregate the metrics. The config file is at [init_database.sql](./config_data/init_database.sql)
-  - we can do other aggregation on top of partial-aggregated data to accelerate the query performance
-  - we can treat the aggregated data as down-sampled data and store it for longer retention
+- A continuous aggregation over metrics. The config file is at
+  [init_database.sql](./config_data/init_database.sql)
+  - we can do other aggregation on top of partial-aggregated data to accelerate
+    the query performance
+  - we can treat the aggregated data as down-sampled data and store it for
+    longer retention
   - we can do conditional aggregation based on some columns
 
-And the data are visualized in Grafana, which is available at [http://localhost:3000](http://localhost:3000) with default username and password as `admin`.
+## Help
+
+1. If you modified component like `vector` in this setup, remember to add
+   `--build <container_name>` to your `docker compose up` to trigger a rebuild.
+2. If you are behind a proxy, add `--build-args <proxy-url>` to `docker compose
+   build` for a full build.
 
-Beside several basic panels, it also has
-- a tree map to show the user-agent corresponding platform that is extracted in pipeline
-- a log details of newest 50 nginx access log
-- two time-series panels shows flow continuous aggregation and conditional aggregation
-- a table shows top 10 slow trace IDs
-- a joined log details shows top 10 slow logs based on provided search strings to trace ID. Both nginx and server sides details are included.
+Feel free to add an issue if you have any question with this demo.