Product image performance

Net-A-Porter are currently looking at dynamic image resizing service by Mr Porter and The Outnet. This project was to check the performance didn't degrade our customer experience.

Once the application has a bunch of PIDs (Product IDs) it will then make 10 requests to origin and the cdn. Each time it will collect how long the response took, generating an average.

Once complete it will save this data and display the results in terminal.

Set up

  nvm use
  npm install

Running

  npm start

Graphing benchmarks

Each time you run a performance test it will save a benchmark inside the folder: benchmarks. If you want to graph an existing benchmark you can:

  BENCHMARK_FILENAME=2016-08-11_14-41-13.json npm start

Configuration

Run the tests only for some of the brands:

    npm start mrp nap

Randomize PID list, default hits whats new pid list, this is more likely to be cached within resize service:

    RANDOM_PIDS=true npm start

Alter the number of PIDS (Product IDs) - default 100:

    NUMBER_OF_PIDS=20 npm start

Alter the number of tests on an image - default 10:

    NUMBER_OF_TESTS=20 npm start

Load local PID list:

    NAP_LOCAL_PIDS=true npm start

You can string multiple local pid lists together:

    NAP_LOCAL_PIDS=true MRP_LOCAL_PIDS=true TON_LOCAL_PIDS=true npm start

If you want to update the lists they need to match the product API PID format:

curl http://lad-api.net-a-porter.com/NAP/GB/20/0/pids?visibility=visible&whatsNew=Now

Then saved in the folder: pids