a igmp query question - TCP-IP

This is a discussion on a igmp query question - TCP-IP ; For igmp v2, it has a seleting query feature. Lower ip box is query, Question is if there are two routers are on the same subnet. I say A_router and B_router (both of igmp enabled) A_router connect to a M'cast ...

+ Reply to Thread
Results 1 to 7 of 7

Thread: a igmp query question

  1. a igmp query question

    For igmp v2, it has a seleting query feature. Lower ip box is query,
    Question is if there are two routers are on the same subnet. I say
    A_router and B_router (both of igmp enabled) A_router connect to a
    M'cast source and B does not. but B has lower ip. Can B become a
    query?



  2. Re: a igmp query question

    On Apr 15, 11:46*pm, aaabb...@hotmail.com wrote:
    > For igmp v2, it has a seleting query feature. Lower ip box is query,
    > Question is if there are two routers are on the same subnet. I say
    > A_router and B_router *(both of igmp enabled) A_router connect to a
    > M'cast source and B does not. but B has lower ip. *Can B become a
    > query?


    The multicast router with the lowest IP address becomes the querier
    (RFC 2236 Section 3).

    So in your case, I am assuming that router B does not connect to this
    particular multicast source, but it is still a multicast enabled
    router. If this is true, then router B becomes the querier for the
    network.

    That works fine. Since all "leave group" reports are multicast to the
    224.0.0.2 "all routers" multicast address, they will reach router A.

    If router B is *not* multicast enabled, then it will not ever send out
    queries. Since router A sees no other queries in this IP subnet,
    router A assumes the role of querier.

    Bert

  3. Re: a igmp query question

    On 4月16日, 上午10时11分, Albert Manfredi wrote:
    > On Apr 15, 11:46 pm, aaabb...@hotmail.com wrote:
    >
    > > For igmp v2, it has a seleting query feature. Lower ip box is query,
    > > Question is if there are two routers are on the same subnet. I say
    > > A_router and B_router (both of igmp enabled) A_router connect to a
    > > M'cast source and B does not. but B has lower ip. Can B become a
    > > query?

    >
    > The multicast router with the lowest IP address becomes the querier
    > (RFC 2236 Section 3).
    >
    > So in your case, I am assuming that router B does not connect to this
    > particular multicast source, but it is still a multicast enabled
    > router. If this is true, then router B becomes the querier for the
    > network.
    >
    > That works fine. Since all "leave group" reports are multicast to the
    > 224.0.0.2 "all routers" multicast address, they will reach router A.
    >
    > If router B is *not* multicast enabled, then it will not ever send out
    > queries. Since router A sees no other queries in this IP subnet,
    > router A assumes the role of querier.
    >
    > Bert


    Thanks, that is why leave report use 224.0.0.2, I think.

  4. Re: a igmp query question

    On 4月16日, 上午10时42分, aaabb...@hotmail.com wrote:
    > On 4月16日, 上午10时11分, Albert Manfredi wrote:
    >
    >
    >
    >
    >
    > > On Apr 15, 11:46 pm, aaabb...@hotmail.com wrote:

    >
    > > > For igmp v2, it has a seleting query feature. Lower ip box is query,
    > > > Question is if there are two routers are on the same subnet. I say
    > > > A_router and B_router (both of igmp enabled) A_router connect to a
    > > > M'cast source and B does not. but B has lower ip. Can B become a
    > > > query?

    >
    > > The multicast router with the lowest IP address becomes the querier
    > > (RFC 2236 Section 3).

    >
    > > So in your case, I am assuming that router B does not connect to this
    > > particular multicast source, but it is still a multicast enabled
    > > router. If this is true, then router B becomes the querier for the
    > > network.

    >
    > > That works fine. Since all "leave group" reports are multicast to the
    > > 224.0.0.2 "all routers" multicast address, they will reach router A.

    >
    > > If router B is *not* multicast enabled, then it will not ever send out
    > > queries. Since router A sees no other queries in this IP subnet,
    > > router A assumes the role of querier.

    >
    > > Bert

    >
    > Thanks, that is why leave report use 224.0.0.2, I think.- 隐藏被引用文字 -
    >
    > - 显示引用的文字 -


    rfc3376 only mention its query 0x11 and report 0x22, and i know
    it support v2 leaving. When v3 host want to leaving group, What
    does it do?
    TIA,
    st

  5. Re: a igmp query question

    On Apr 18, 1:39*pm, aaabb...@hotmail.com wrote:

    > rfc3376 only mention its query 0x11 and report 0x22, and i know
    > it support v2 leaving. When v3 host want to leaving group, What
    > does it do?


    First of all, IGMPv3, like IGMPv2, defaults always to the oldest
    version of IGMP that the router notices existing in the network. Even
    though all queries use Type 0x11, Section 7.1 of RFC 3376 explains how
    the version of IGMP can be determined from a query message. In short,
    it is determined by the existence of a max response time variable, and
    by the length of teh query message.

    Secondly, IGMPv3 reports are always sent to 224.0.0.22 (Section
    4.2.14) and use an IGMP Type of 0x22, so they can be distinguished
    from IGMPv1 or v2 reports.

    When in IGMPv3 mode, the type of IGMP report that indicates "leave
    group" is just another IGMP report, instead of being a separate type
    of message as it was in IGMPv2.

    So with IGMPv3, if a host wants to indicate that it has dropped
    membership, it sends a report (to address 224.0.0.22) which indicates
    membership in the specified IP multicast group, to INCLUDE the source
    addresses listed in the IGMPv3 report, and then lists no source
    addresses are listed. From this, the router can conclude that this
    particular host does not want to receive any multicasts in that
    multicast group. This is in Section 5.1.

    Bert

  6. Re: a igmp query question

    On 4月18日, 下午1时16分, Albert Manfredi wrote:
    > On Apr 18, 1:39 pm, aaabb...@hotmail.com wrote:
    >
    > > rfc3376 only mention its query 0x11 and report 0x22, and i know
    > > it support v2 leaving. When v3 host want to leaving group, What
    > > does it do?

    >
    > First of all, IGMPv3, like IGMPv2, defaults always to the oldest
    > version of IGMP that the router notices existing in the network. Even
    > though all queries use Type 0x11, Section 7.1 of RFC 3376 explains how
    > the version of IGMP can be determined from a query message. In short,
    > it is determined by the existence of a max response time variable, and
    > by the length of teh query message.
    >
    > Secondly, IGMPv3 reports are always sent to 224.0.0.22 (Section
    > 4.2.14) and use an IGMP Type of 0x22, so they can be distinguished
    > from IGMPv1 or v2 reports.
    >
    > When in IGMPv3 mode, the type of IGMP report that indicates "leave
    > group" is just another IGMP report, instead of being a separate type
    > of message as it was in IGMPv2.
    >
    > So with IGMPv3, if a host wants to indicate that it has dropped
    > membership, it sends a report (to address 224.0.0.22) which indicates
    > membership in the specified IP multicast group, to INCLUDE the source
    > addresses listed in the IGMPv3 report, and then lists no source
    > addresses are listed. From this, the router can conclude that this
    > particular host does not want to receive any multicasts in that
    > multicast group. This is in Section 5.1.
    >
    > Bert


    Thanks,
    if iptv use these protocols, the delay is too much i thik. How to make
    it less than100ms from one channle to another?

    HIA,
    st


  7. Re: a igmp query question

    On Apr 18, 11:44*pm, aaabb...@hotmail.com wrote:

    > if iptv use these protocols, the delay is too much i thik. How to make
    > it *less than100ms from one channle to another?


    You can tune the timers, to reduce the delay from what the default
    values are.

    Digital TV, even the non-IP kind, creates longer channels switching
    delays than analog TV. One of the main reasons why is that you cannot
    sync up to the new program stream until you reach an MPEG I frame. And
    much of the efficiency of MPEG is achieved by spreading out I frames
    as much as possible. So no matter what, people will be used to
    switching times approaching 1 second, if not more, in any form of DTV.

    Another trick in IPTV will be to have the most popular programs always
    available between the backbone network and the edge routers, even if
    there is no current membership in the edge networks themselves. So
    when an end system joins an IP multicast group, the edge router will
    immediately have the packets of that multicast group available,
    without having to search upstream.

    The default values of the IGMP timers probably don't apply to IPTV
    networks. They are chosen to try to reduce network load created by
    IGMP. An IPTV network will require plenty of over-provisioning anyway,
    just to achieve good QoS, so I doubt a little more IGMP loading would
    be such a big problem.

    Bert

+ Reply to Thread