aludwiko

It doesn’t matter if you are from JVM, .Net, PHP or else-world. If you need to test the performance - it will be a challenging task, especially nowadays with microservices architectures, clusters and very complex systems. I would like to address the most common pitfalls in this area. Share my experience gained through demanding experiments and quite often frustrating failures. Although most of the examples come from the JVM world, the aim of this presentation is to show some universal problems, laws and best practices that rule this very specific kind of testing.

aludwiko

val eventJournal  = PersistenceQuery(system).readJournalFor[CassandraReadJournal](CassandraReadJournal.Identifier)

eventJournal
  .eventsByPersistenceId(persistenceId, startingSequenceNr, Long.MaxValue)
  .map {
    case EventEnvelope(_, _, sequenceNr, event: DomainEvent) => {
      (sequenceNr, event)
    }
  }
  .via(sendEventToKafka)

aludwiko

       Value     Percentile TotalCount 1/(1-Percentile)
           0 0.000000000000        970           1.00
          63 0.500000000000      64196           2.00
         127 0.750000000000     100000           4.00
         127 0.875000000000     100000           8.00
         127 0.937500000000     100000          16.00
         127 0.968750000000     100000          32.00
         127 0.984375000000     100000          64.00
         127 0.992187500000     100000         128.00
         127 0.996093750000     100000         256.00

aludwiko

def printHdrHistogram(times: Seq[Int]) = {
 val histogram = new Histogram(1000000, 0)
 times.map(time => histogram.recordValue(time))
 histogram.outputPercentileDistribution(System.out, 1, 1.0)
}

aludwiko

-----------1
75: 75.000000
95: 95.000000
99: 98.000000
99.9: 1000.000000
max: 1000
-----------2
75: 75.000000
95: 94.000000
99: 98.000000

aludwiko

object DropWizardTest extends App {
  val random = Random

  val times =
    //normal behavior
    (1 to 100000).map(_ => random.nextInt(100)) ++
      //standard peacks
      (1 to 100).map(_ => 1000) ++
      //very high peacks
      (1 to 10).map(_ => 10000)

aludwiko

Source(1 to 10)
 .map(_ / 0)
 .runWith(Sink.ignore)

aludwiko

Source(1 to 10)
 .flatMapConcat { i =>
   Source.single(i)
    .map(toKafkaRecord)
    .via(Producer.flow(producerSettings))
 }
 .runForeach(i => println(s"${i.message.passThrough} sent to kafka"))

aludwiko

Source(1 to 10)
 .map(toKafkaRecord)
 .via(Producer.flow(producerSettings))
 .runForeach(i => println(s"${i.message.passThrough} sent to kafka"))

aludwiko

Source(1 to 10)
 .via(nonLinearFlow)
 .collect {
   case (Success(a), Success(b)) => (a, b)
 }
 .runForeach(println)