Scribd Logo
Upload

Welcome to Scribd!

  • 88 views

    Load Balancing

    Uploaded by

    Vivi Vera
    Load balancing is a technique used to distribute service requests across multiple computing resources to maximize throughput, increase utilization, and prevent overload. It can be implemented in hardware using load balancers or in software. Common load balancing algorithms distribute requests in a round-robin fashion or to resources with the fewest connections or fastest response times. Load balancing improves reliability, provides redundancy, and increases tolerance to failures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Load Balancing

    Uploaded by

    Vivi Vera
    88 views25 pages
    Load balancing is a technique used to distribute service requests across multiple computing resources to maximize throughput, increase utilization, and prevent overload. It can be implemented in hardware using load balancers or in software. Common load balancing algorithms distribute requests in a round-robin fashion or to resources with the fewest connections or fastest response times. Load balancing improves reliability, provides redundancy, and increases tolerance to failures.

    Original Description:

    Balanceador de carga

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Load balancing is a technique used to distribute service requests across multiple computing resources to maximize throughput, increase utilization, and prevent overload. It can be implemented in hardware using load balancers or in software. Common load balancing algorithms distribute requests in a round-robin fashion or to resources with the fewest connections or fastest response times. Load balancing improves reliability, provides redundancy, and increases tolerance to failures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    88 views25 pages

    Load Balancing

    Uploaded by

    Vivi Vera
    Load balancing is a technique used to distribute service requests across multiple computing resources to maximize throughput, increase utilization, and prevent overload. It can be implemented in hardware using load balancers or in software. Common load balancing algorithms distribute requests in a round-robin fashion or to resources with the fewest connections or fastest response times. Load balancing improves reliability, provides redundancy, and increases tolerance to failures.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 25

    Load Balancing

    Technology used to distribute service requests to

    resources.
    Load Balancing is an optimization technique in order
    to increase utilization, yield, lowest latency,
    response time reduction and prevent the system
    overload.
    This can be implemented in hardware e.g.: F5 BigIP
    Servers
    And can also be implemented in software e.g.:
    Apache mod_proxy_balancer extension

    Load Balacing
    The following network resources can be load

    balanced:
    Network Interfaces and services such as DNS, FTP,

    and HTTP
    Connections (using intelligent switches)
    Storage Resources
    Access to application instances

    Load Balancing
    Without Loud Balancing would be very difficult to

    manage the cloud (in a strand of requests VS


    resources).
    Load balancing provides redundancy needed to make
    an unreliable system into trusted through the
    redirection of resources by the machines.
    In addition a grouping with a failover mechanism

    provides an increased tolerance to failures.

    Load Balancing
    Operating principles
    In simpler load-balancing mechanisms:
    The Loud Balancing is listening on a network port for
    requests for services
    When a client makes a request, it uses a scheduling
    algorithm to allocate where the request shall be sent

    Load Balancing
    Typical algorithms of timing:
    Round-Robin
    Weighted Round Robin
    Dynamic Round Robin
    Fastest response Time
    Least Connections

    Load Balancing - Algorithms


    Fastest Response Time
    The server is chosen based on the faster response time of a group of servers.
    This method can be particularly useful in environments where the servers are
    distributed in different logical networks.

    Its possible only because of a good


    load balancing mechanism
    /*Just for fun*/

    Internet

    WAN

    [longman] Mechanism:
    um sistema que se destina a
    alcanar algo ou resolver
    um problema

    Load Balancing - Mechanisms


    Firewall

    Balanceamento de carga executadoSwitch


    pelo balanceadores de carga,

    usando algoritmos de agendamento.


    Load balancer

    LAN

    Switch

    Server 1

    Server 2

    Server 3

    Load Balancing - Mechanisms


    Balanceamento de carga executado pelo balanceadores de carga, usando

    algoritmos de agendamento.

    hardware-based
    service-based

    Load Balancing - Mechanisms


    exemplo do balanceador de carga baseada em hardware:

    Load Balancing - Mechanisms


    service-based load balancer example:

    Load Balancing - Mechanisms

    Balanceamento de carga executado pelo balanceadores de carga,


    usando algoritmos de agendamento.

    Algoritmos de agendamento :

    Esttica: no h troca de informaes entre o balanceador de carga e


    cada n de servidor, por isso tem um limite na conduo correta do
    balanceamento.

    Dinmica: rotear solicitaes do cliente de acordo com as informaes


    dos ns do servidor.

    Load Balancing - Algorithms


    Algoritmos de agendamento estticos incluem (normalmente):

    Round-Robin: atribui os pedidos para os servidores do modo Round-Robin


    Least-Connection: encaminhar as solicitaes para o servidor que tenha o
    mnimo de conexes
    Weighted Round-Robin and Weighted Least-Connection: D o peso de cada
    n de servidor

    Algoritmos de agendamento dinmicos incluem (normalmente):

    Agent-based Sophisticated and Scalable Scheduling Method (ABSS)


    Agent-based Sophisticated Scheduling Mechanism by Using Memory
    Utilization (ASSUM)
    Dynamic Round-Robin

    Load Balancing - Algorithms


    Round-Robin
    Um dos mtodos mais simples para distribuir solicitaes do cliente

    atravs de um grupo de servidores.

    Round-Robin manner: O balanceador de carga encaminha as solicitaes


    de cliente para cada servidor vez. Quando atingir o final da lista, o
    balanceador de carga volta ao comeo.

    Classificao do mtodo Round-Robin :

    Static
    Weighted
    Dynamic

    Load Balancing - Algorithms


    Static Round-Robin
    O balanceador de carga percorre uma fila circular

    representativa de servidores, divide e envia as


    solicitaes do cliente para cada servidor antes de
    chegar ao incio da fila para comear de novo.
    Funciona bem quando todos os servidores tm a

    mesma capacidade.

    Load Balancing - Algorithms


    Static Round-Robin
    Client
    A, B, C, A, B, C, and so on

    Load
    balancer

    Server
    A

    Server
    B

    Server
    C

    Load Balancing - Algorithms


    Weighted Round-Robin

    atribuda um peso para cada servidor com base em critrios escolhidos


    pelo administrador.

    O critrio mais comumente usado a capacidade de manipulao de


    trfego do servidor. O balanceador de carga faz uma solicitao HTTP e a
    medida de quo rpida que um servidor responde a ele.

    Quanto maior o peso, maior a proporo de solicitaes de clientes o


    servidor recebe.

    Load Balancing - Algorithms


    Weighted Round-Robin
    Client
    A, B, B, C, A, B, B, C, and so on

    Load
    balancer

    Server
    A

    Server
    B

    Server
    C

    Load Balancing - Algorithms


    Dynamic Round-Robin

    Um peso atribudo a cada servidor dinamicamente, com base em dados


    em tempo real sobre o estado atual do servidor, portanto, o peso
    alterado continuamente.

    Um mtodo dinmico de balanceamento de carga, distribuindo os


    pedidos com base em vrios aspetos da anlise de servidor em tempo real,
    tais como o nmero atual de conexes por servidor ou o mais rpido
    tempo de resposta do servidor.

    Load Balancing - Algorithms


    Least Connections
    With this method the system consider connections with a server that has the
    least number of current connections.
    Least Connections works best in environments where the servers and/or
    equipment have similar capabilities.

    Concluding, this algorithm only keeps track of the number of links attached to
    each server.
    Selecting the one with the fewest

    Load Balancing - Algorithms


    Observed
    This method uses a combination of logic used in the Least connections
    algorithms and Fastest response.
    In this method the servers that have a better balance: fewer connections and
    faster response time.
    Receive a larger proportion of links

    Load Balancing - Algorithms


    Predictive
    The Predictive uses the classification method used by the algorithm Observed.
    However, with this algorithm it is possible to analyze the trend of ranking over
    time:
    Determine whether the performance of one or more servers is improving or getting

    worse.

    The ranked servers with a better performance rating, rather than worsen,
    receive a higher proportion of links.

    Load Balancing - Algorithms


    Random
    This method of load balancing distributes randomly the load between available
    servers.
    Chooses random numbers (assigned to different servers) and sends the current
    connection to the chosen.

    This method is available in many implementations of load balancing.


    However, his utility is questionable.

    Load Balancing - Conclusions


    Having examined the different balancing algorithms we can conclude:
    Load-balancing systems more sophisticated manage differently the

    workload requested and/or allocated by customers based on different


    factors:

    Response time
    Waiting lists
    Connection latency and capacity

    Load Balancing
    Thank you
    Obrigado

    You might also like