Computer Network - Queueing Theory

기본

    Quantitative

        Capacity

        Number of Transmission Links ( Routers )

            ex) Network 를 보안에 따라 나누기

        Buffer Sizes

            버퍼크기 = RTT * Link_Capacity / root(number of TCP flow)

        Size of the Switching Fabric

    Metrics

        Call Blocking => Capacity 보다 초과시 블럭

        Packet loss Probability => Capacity, Buffer Size 

        Moments and Distribution Delay

            Exp(D), Variation(D), Pr{D>x} ( Guarantee )

        Throughput

    Above is Statistical(확률적) 

        as Deterministic Quantity make Under Utilization

        don't know parameters that make randomness

            Arrival Packet time, Holding Time, Pack Size, Transmission Facility

        => Queueing Theory

            이런 랜덤 가득한 네트워크의 성능을 분석하고 만들기 위해 필요

    Queue is Variable

        Lots of Job + Limited Resources => Make Queue

        Delay Delay = Processing + Queue + Transmission + Propagation

        Queue => 나머진 다 상수인데 얘만 가변적임 => Queue Theory

Theory of Queue

    Server => S

    Buffer => B

        at least 1 Packet Length as It should store for Transmission

    S < B imply Packet Loss Probability => 1/(S*B)

    ...

    왜 1에 가까울수록 queue 확률이 커지냐를 수학적으로 증명하는건데 나중에