Design for The Benchmark

Models:

• Question belongs_to User
• Answer belongs_to User
• Answer belongs_to Question

Note:

• use seed to mock mass of data
• provide API to get questions/answers/users
• use sleep to simulate calculate in real situation
• request 10 API to zoom in the request time
• change data between request to simulate data change in real situation
• Change/Request as an indicator for application data change frequency

How to Benchmark ams cache

1. comment cached: true in every Serializer file
2. run bin/rake benchmark:ams:static to get the uncache data for reference group
3. uncomment cached: true
4. change Request Count(r) and Change Count(c), and run bin/rake "benchmark:ams:dynamic[r, c]"
5. rm tmp/cache directory and run Step 4

How to Benchmark jbuilder cache

1. comment json.cache! in every .jbuilder file
2. run bin/rake benchmark:jbuilder:static to get the uncache data for reference group
3. uncomment json.cache!
4. change Request Count(r) and Change Count(c), and run bin/rake "benchmark:jbuilder:dynamic[r, c]"
5. rm tmp/cache directory and run Step 4

Benchmark Result

benchmark result

ams cache result

jbuilder cache result

notes:

• x is Change/Request, Request total is 10
• y is No Cache Request Time / Cached Request Time