How to deal with late data | Event vs Processing Time | Stream Processing

Поделиться
HTML-код
  • Опубликовано: 8 ноя 2024

Комментарии • 9

  • @bjugdbjk
    @bjugdbjk 2 года назад

    Good to see u back buddy !!

  • @JardaniJovonovich192
    @JardaniJovonovich192 2 года назад

    have you used Kafka streams in production? If yes, what tool have you used to load test such application?

    • @irtizahafiz
      @irtizahafiz  2 года назад

      Hi! I work at a Product team, not an Infra one. So unfortunately I am not really sure what tools are used. They are abstracted away from us.

  • @anidea8012
    @anidea8012 Год назад

    I'm currently working on a project in which I'm getting event time > process time; for example, in my database, the process time happened in April and my event time happened in May. How is this possible? Please help me. Or is this any kind of cyberattack?

    • @irtizahafiz
      @irtizahafiz  11 месяцев назад

      Can there be errors when recording the event time? Maybe the events were sending the wrong timestamp to begin with? Or vice-versa, maybe the process time is recorded incorrectly.

  • @kushangowda
    @kushangowda Год назад

    Why are you over engineering
    Why don’t you use timestamp to identify?

    • @irtizahafiz
      @irtizahafiz  11 месяцев назад +1

      If you are doing window-based aggregations, and your event arrives beyond the window "grace period", you will either ignore the data or need to recompute.

    • @kavinkumar2633
      @kavinkumar2633 8 месяцев назад

      For systems like ad aggregator, it makes sense to ignore the late data ,correct?@@irtizahafiz