TRDL is a web service that returns the value
42through a public API, the value is fetched from a database.
The requirements on TRDL are that the production system is a high availability service with a strict service level agreement (several "9" s uptime).
The system consists of an HTTP server with some code running inside or behind it, and a database. It responds to a GET request as in the example.
sequenceDiagram
participant User
participant HTTP server
participant Application
participant Database
User-->>HTTP server: get value
HTTP server-->>Application: get value
Application-->>Database: get value
Database-->>Application: return value
Application-->>HTTP server: return value
HTTP server-->>User: return value
- Install
direnv. - Configure
direnvto load.env. - Install Docker Compose.
- Create an
.envfile with the environment variable setting the superuser password for PostgreSQL. - Create and start containers.
$ echo "POSTGRES_PASSWORD='$(openssl rand -base64 16)'" > .env
$ direnv allow
direnv: export +POSTGRES_PASSWORD
$ docker compose up --wait
[+] Running 18/18
✔ web 8 layers [⣿⣿⣿⣿⣿⣿⣿⣿] 0B/0B Pulled 10.8s
# ...
✔ db 8 layers [⣿⣿⣿⣿⣿⣿⣿⣿] 0B/0B Pulled 14.4s
# ...
[+] Building 43.2s (13/13) FINISHED
# ...
[+] Running 8/8
✔ Network trdl_db Created 0.2s
✔ Network trdl_app Created 0.1s
✔ Volume "trdl_pgdata" Created 0.0s
✔ Container trdl-db-1 Healthy 4.5s
✔ Container trdl-app-3 Healthy 32.9s
✔ Container trdl-app-1 Healthy 32.9s
✔ Container trdl-app-2 Healthy 32.9s
✔ Container trdl-web-1 Healthy 4.1s
$ docker-compose run app flask init-db
[+] Building 0.0s (0/0)
[+] Creating 3/3
✔ Network trdl_app Created 0.1s
✔ Network trdl_db Created 0.1s
✔ Container trdl-db-1 Created 0.1s
[+] Running 1/1
✔ Container trdl-db-1 Started 0.3s
[+] Building 0.0s (0/0)Here is a sample interaction with TRDL:
$ curl http://127.0.0.1:8080/v1/value
42(...) once the application is deployed it will be used by thousands of customers from day 1.
- Install the application.
- Install
ab- an Apache HTTP server benchmarking tool. - Run the benchmark.
$ ab -n 2000 -c 1 http://127.0.0.1:8080/v1/value
This is ApacheBench, Version 2.3 <$Revision: 1903618 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking 127.0.0.1 (be patient)
Completed 200 requests
Completed 400 requests
Completed 600 requests
Completed 800 requests
Completed 1000 requests
Completed 1200 requests
Completed 1400 requests
Completed 1600 requests
Completed 1800 requests
Completed 2000 requests
Finished 2000 requests
Server Software: nginx/1.25.1
Server Hostname: 127.0.0.1
Server Port: 8080
Document Path: /v1/value
Document Length: 2 bytes
Concurrency Level: 1
Time taken for tests: 16.168 seconds
Complete requests: 2000
Failed requests: 0
Total transferred: 316000 bytes
HTML transferred: 4000 bytes
Requests per second: 123.70 [#/sec] (mean)
Time per request: 8.084 [ms] (mean)
Time per request: 8.084 [ms] (mean, across all concurrent requests)
Transfer rate: 19.09 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.4 0 16
Processing: 4 8 12.5 6 323
Waiting: 4 7 12.4 6 322
Total: 4 8 12.7 7 339
Percentage of the requests served within a certain time (ms)
50% 7
66% 8
75% 8
80% 9
90% 10
95% 12
98% 14
99% 21
100% 339 (longest request)SLA level of 99.9 % uptime/availability results in the following periods of allowed downtime/unavailability:
- Daily: 1m 26s
- Weekly: 10m 4.8s
- Monthly: 43m 28s
- Quarterly: 2h 10m 24s
- Yearly: 8h 41m 38s
The quote below is from Example SLO Document by Steven Thurgood. The document is licensed under the CC BY-NC-ND 4.0 license. © 2018 Google, Inc.
The proportion of successful requests, as measured from the load balancer metrics.
Any HTTP status other than 500–599 is considered successful.
- 99%
The proportion of sufficiently fast requests, as measured from the load balancer metrics.
Sufficiently fast is defined as < 200 ms, or < 1,000 ms.
- 90% of requests < 200 ms
- 99% of requests < 1,000 ms