HS algorithm for Leader Election in Distributed Systems

Поділитися
Вставка
  • Опубліковано 31 жов 2024

КОМЕНТАРІ • 12

  • @shishirchaurasiya7374
    @shishirchaurasiya7374 Рік тому +1

    Amazing Amazing Completely amazing video Arpit sir
    I am in literally in love in your teaching style

  • @myslates2854
    @myslates2854 2 місяці тому

    Hi Arpit, Can you make the column of text shorter in width as some of the diagrams are hidden with your camera shot. It would be good to see all the text and camera shot without overlapping.

  • @shantanutripathi
    @shantanutripathi 2 роки тому

    Correct me if I'm wrong. For Implementation, the message which contains must contains one clone to hops (say count) and we should decrease that.. Otherwise there is no point of knowing how many hops, when going in reverse direction.

    • @vatsalmehta6644
      @vatsalmehta6644 2 роки тому

      while coming in reverse direction we have uid right which originally sent this message. Once node with that uid gets back it will terminate and won't forward(in reverse direction) again. If such message received from both end then will be proceeding for next phase. This is what I understood :). @Arpit may be you can confirm :D

    • @balakrishnan3725
      @balakrishnan3725 2 місяці тому

      While coming in reverse direction, there should be an additional attribute which should say that this is in reverse direction. Or even the existing "direction" attribute should hold the value as "reverse_left" or "reverse_right". i.e., from the direction value the nodes should able to identify that this message is reverse for the forward message sent before. Other than that the hops count value is not needed in case of reverse. Once the initiator receives the message then the reverting stops. This is my understanding.

  • @RafikYacoub
    @RafikYacoub Рік тому

    Is this algorithm any like bully or ring algorithms?

  • @ashleymavericks
    @ashleymavericks 2 роки тому

    If we are talking about knowing the hop count, doesn't it directly means that we need to know the total number of nodes in advance?

    • @AsliEngineering
      @AsliEngineering  2 роки тому +1

      Nope. It just increases exponentially. Check the termination phase of the algorithm.

  • @Aditya-us5gj
    @Aditya-us5gj 2 роки тому

    In LCR algo, the whole probem was that all the nodes were not aware which node crashed and everyone wanted to be leader and initiated the election. This casuing every node sending it's UUID to every next node. This Algo looks good but again my confustion is same, the same scenario will arise here as well. How a node will decide that I will be sending(initiating election) the bi-directional message ? While 7 is sending message to 3 and 9, won't 3 and 9 individually starting sending thier UUID and similarly all the nodes will do this ? Phases will proceed and collection of nodes will increase(going from local to global maxima) if nodes have patience and knows someone has started election. Otherwsie anyways all will be sending the messages bi-directionally

    • @AsliEngineering
      @AsliEngineering  2 роки тому

      Yes. Everyone might but still everyone converges to the same value.

    • @Aditya-us5gj
      @Aditya-us5gj 2 роки тому

      @@AsliEngineering oh, so it's kind of there will be bi-directional messages flooding throughout the nodes but at the end looking at the conditions the algorithm follows, the result will boil down as expected.

  • @molamola8305
    @molamola8305 6 місяців тому

    Why can't we simply go through all nodes and find the node with maximum UID, isn't that simple bfs?