✂️ Shorty

An URL shortening service

How it works

• AWS API Gateway takes requests globally and passes them to AWS Lambda
• AWS Lambda is an auto-scalable serverless solution from AWS. In this case we use containerized approach instead of just uploading the code.
• AWS S3 is used as a storage of the URLs. Each URL is stored as an object with a short random name, the URL slug. Also we specify the target URL in a special header of the object. Once somebody is trying to open the referenced object — they will be redirected to the target URL. This is a standard S3 mechanism
• URLs older than 90 days will be removed automatically with another S3 mechanism

Availability — 99.95%

• AWS Lambda — 99.95%
• AWS S3 — 99.99%
• AWS API Gateway — 99.95%

How it performs — 297.84ms avg response time

running (40.0s), 00000/10000 VUs, 7140 complete and 3173 interrupted iterations
default ✓ [======================================] 10000 VUs  10s

     data_received..............: 49 MB   1.2 MB/s
     data_sent..................: 5.7 MB  142 kB/s
     http_req_blocked...........: avg=19.77s   min=0s       med=20.29s   max=39.82s   p(90)=31.94s  p(95)=33.74s
     http_req_connecting........: avg=5.35s    min=0s       med=4.49s    max=27.42s   p(90)=11.36s  p(95)=11.47s
     http_req_duration..........: avg=2.72s    min=0s       med=297.84ms max=38.36s   p(90)=9.66s   p(95)=14.6s
     http_req_failed............: 100.00% ✓ 7140       ✗ 0
     http_req_receiving.........: avg=329.85ms min=0s       med=29µs     max=29.42s   p(90)=53.56ms p(95)=1.7s
     http_req_sending...........: avg=356.35µs min=0s       med=57µs     max=171.52ms p(90)=114µs   p(95)=196µs
     http_req_tls_handshaking...: avg=14.48s   min=0s       med=14.61s   max=36.73s   p(90)=28.35s  p(95)=31.16s
     http_req_waiting...........: avg=2.39s    min=0s       med=244.6ms  max=38.36s   p(90)=8.66s   p(95)=13.93s
     http_reqs..................: 7140    178.325517/s
     iteration_duration.........: avg=23.04s   min=348.96ms med=23.41s   max=39.92s   p(90)=33.06s  p(95)=35.86s
     iterations.................: 7140    178.325517/s
     vus........................: 3232    min=3232     max=10000
     vus_max....................: 10000   min=10000    max=10000

💸 Costs

• do the estimation https://calculator.aws/#/addService

🔐 Security improvements

• https://aws.amazon.com/blogs/apn/simplify-and-secure-terraform-workflows-on-aws-with-dynamic-provider-credentials/

Improvements backlog

• introduce environments, now it is just prod
• proper versioning for CD, now it is just hardcoded
• introduce variables in *.tf where it makes sense, now many things are just hardcoded, because it is a POC

Useful commands

# run the load test to see how the system performs
k6 run loadtest.js

# to build the container and test it locally
npm run serve
# would run a query against serving container
npm run test
# to deploy the latest changes to the repository
npm deploy

# and also some other commands
brew install awscli
aws configure
aws sts get-caller-identity
aws sts get-session-token