General Ideas

Timeout exercise :

• In 10% of all cases the response arrives in less than 10ms.
• In 50% of all cases the response arrives in less than 20ms.
• In 90% of all cases the response arrives in less than 25ms.
• In 99% of all cases the response arrives in less than 50ms.
• In 99.9% of all cases the response arrives in less than 120ms.
• In 99.99% of all cases the response arrives in less than 470ms.
• In 99.999% of all cases the response arrives in less than 1340ms.
• The largest latency ever observed was 3 minutes and 16 seconds.

• LITTLE’S LAW = L = avR W L - average numer of requests served at the same time
  • avr - average rate Requests arrive to system
  • W - time to process a request
  • "if there’s on average one request per minute and each request takes 100 ms to process, the average number of concurrent requests will be approximately 0.0017. In other words,there is no need to worry about scaling out for this application."
  • If one node can process 70 requests, how many nodes do you need? (two variants one prepared and one random, generate requests and measure!! - use akka for simulation.!!)

Later

• testing actors
  • bigger example with tests

• routing actors

• TRANSAKCJE - transakcja jako aktor ze stanem, maszyna stanów?

Performance

• https://scalameter.github.io/