Security

NAVIGATION
CATEGORIES
REFERRENCE
LINKS
  • unswitched behavior of a switched network...

    13 answers - 1624 bytes - related search similar search Add To My Delicious Add To My Stumble Upon Add To My Google Mark Add To My Facebook Add To My Digg Add To My Reddit

    Greetings,
    I've got a situation here that I can't quite figure out. It is well
    known that it is possible to cause a switched network to act like an
    unswitched network by flooding the CAM table. There are countless tools
    and documents out there that cover the offensive and defensive measures
    related to this issue.
    While this isn't Cisco's official documentation on this issue,
    http://xrl.us/r8k7 says:
    "Content-addressable memory (CAM) overflow: A CAM table is used to
    determine where to direct incoming frames depending on which port the
    incoming MAC address came from. When the CAM receives a frame with an
    unknown destination, the proper procedure is to flood frames within
    the acceptable Layer 2 domain (the proper VLAN). Hardware and
    software tools are available (some for free), that can flood a switch
    with MAC addresses. the CAM table limit is exceeded, switches
    behave differently depending on the brand of the switch."
    My question is, has anyone seen a situation where the same broadcast
    behavior occurs, but the CAM table itself is not overloaded and there is
    no good reason for entries to be expiring? Furthermore, even if the
    entries were expired, has anyone encountered situations (malicious or
    otherwise), where a given port will receive traffic outside of its own
    L2?
    Thanks,
    -jon
    This List Sponsored by: Cenzic
    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.
  • No.1 | | 1989 bytes | |

    10/13/06, Jon Hart <jhart (AT) spoofed (DOT) orgwrote:
    Greetings,

    I've got a situation here that I can't quite figure out. It is well
    known that it is possible to cause a switched network to act like an
    unswitched network by flooding the CAM table. There are countless tools
    and documents out there that cover the offensive and defensive measures
    related to this issue.

    While this isn't Cisco's official documentation on this issue,
    http://xrl.us/r8k7 says:

    "Content-addressable memory (CAM) overflow: A CAM table is used to
    determine where to direct incoming frames depending on which port the
    incoming MAC address came from. When the CAM receives a frame with an
    unknown destination, the proper procedure is to flood frames within
    the acceptable Layer 2 domain (the proper VLAN). Hardware and
    software tools are available (some for free), that can flood a switch
    with MAC addresses. the CAM table limit is exceeded, switches
    behave differently depending on the brand of the switch."

    My question is, has anyone seen a situation where the same broadcast
    behavior occurs, but the CAM table itself is not overloaded and there is
    no good reason for entries to be expiring? Furthermore, even if the
    entries were expired, has anyone encountered situations (malicious or
    otherwise), where a given port will receive traffic outside of its own
    L2?

    Thanks,

    -jon

    Some router have an option of dumping all traffic to a give port, so
    if you are connected to the right router port you will see everything
    as if it was a hub. At least I already saw a router configured that
    way, that port that was connected to a computer that was dedicated to
    run snort.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.2 | | 2280 bytes | |

    Is all traffic being "broadcasted"? Can you narrow it down to a
    specific host that's common to all of the traffic, perhaps a gateway
    device? If you're doing multicasting on a gateway device
    (multicasting using unicast addressing), you would get the type of
    behavior that you're describing. I've seen this exact situation
    before, actually.

    BN

    10/13/06, Jon Hart <jhart (AT) spoofed (DOT) orgwrote:
    Greetings,

    I've got a situation here that I can't quite figure out. It is well
    known that it is possible to cause a switched network to act like an
    unswitched network by flooding the CAM table. There are countless tools
    and documents out there that cover the offensive and defensive measures
    related to this issue.

    While this isn't Cisco's official documentation on this issue,
    http://xrl.us/r8k7 says:

    "Content-addressable memory (CAM) overflow: A CAM table is used to
    determine where to direct incoming frames depending on which port the
    incoming MAC address came from. When the CAM receives a frame with an
    unknown destination, the proper procedure is to flood frames within
    the acceptable Layer 2 domain (the proper VLAN). Hardware and
    software tools are available (some for free), that can flood a switch
    with MAC addresses. the CAM table limit is exceeded, switches
    behave differently depending on the brand of the switch."

    My question is, has anyone seen a situation where the same broadcast
    behavior occurs, but the CAM table itself is not overloaded and there is
    no good reason for entries to be expiring? Furthermore, even if the
    entries were expired, has anyone encountered situations (malicious or
    otherwise), where a given port will receive traffic outside of its own
    L2?

    Thanks,

    -jon
    --

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    --

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.3 | | 2667 bytes | |

    Friday 13 2006 18:32, Jon Hart wrote:

    hi jon,

    i had a similar situation in a switched environment. certain frames to a
    particular server where to be seen at every port on all switches within the
    same vlan. the reason was that the server was attached with several cards for
    loadbalancing. arp request for the virtual address where answered by each
    server card but when the client send ip packets using the learned virtual mac
    the server cards replied using their physical address which is stupid since
    the vmac was never used as a source and so it could not be learned by the
    switches. as a result frames that had the vmac as a destination where always
    floodedalso a nice example on how to turn expensive network equipment into
    a hub :)

    regards

    jan

    Greetings,

    I've got a situation here that I can't quite figure out. It is well
    known that it is possible to cause a switched network to act like an
    unswitched network by flooding the CAM table. There are countless tools
    and documents out there that cover the offensive and defensive measures
    related to this issue.

    While this isn't Cisco's official documentation on this issue,
    http://xrl.us/r8k7 says:

    "Content-addressable memory (CAM) overflow: A CAM table is used to
    determine where to direct incoming frames depending on which port the
    incoming MAC address came from. When the CAM receives a frame with an
    unknown destination, the proper procedure is to flood frames within
    the acceptable Layer 2 domain (the proper VLAN). Hardware and
    software tools are available (some for free), that can flood a switch
    with MAC addresses. the CAM table limit is exceeded, switches
    behave differently depending on the brand of the switch."

    My question is, has anyone seen a situation where the same broadcast
    behavior occurs, but the CAM table itself is not overloaded and there is
    no good reason for entries to be expiring? Furthermore, even if the
    entries were expired, has anyone encountered situations (malicious or
    otherwise), where a given port will receive traffic outside of its own
    L2?

    Thanks,

    -jon
    --

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    >W


    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.4 | | 623 bytes | |

    >
    Some router have an option of dumping all traffic to a give port, so
    if you are connected to the right router port you will see everything
    as if it was a hub. At least I already saw a router configured that
    way, that port that was connected to a computer that was dedicated to
    run snort.
    Just to clarify, I'm pretty sure you're talking about switches that have
    a "mirror" port.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.5 | | 2561 bytes | |

    I can think if a couple of possibilities. 1) This is
    broadcast/multicast traffic. 2) The mac addresses are unknown to the
    switch (So it will flood to find them.) 3) The port could be a trunk or
    a mirror of a trunk.

    Buz

    Krugger wrote:
    10/13/06, Jon Hart <jhart (AT) spoofed (DOT) orgwrote:
    >Greetings,
    >>

    >I've got a situation here that I can't quite figure out. It is well
    >known that it is possible to cause a switched network to act like an
    >unswitched network by flooding the CAM table. There are countless tools
    >and documents out there that cover the offensive and defensive measures
    >related to this issue.
    >>

    >While this isn't Cisco's official documentation on this issue,
    >http://xrl.us/r8k7 says:
    >>

    >"Content-addressable memory (CAM) overflow: A CAM table is used to
    >determine where to direct incoming frames depending on which port the
    >incoming MAC address came from. When the CAM receives a frame with an
    >unknown destination, the proper procedure is to flood frames within
    >the acceptable Layer 2 domain (the proper VLAN). Hardware and
    >software tools are available (some for free), that can flood a switch
    >with MAC addresses. the CAM table limit is exceeded, switches
    >behave differently depending on the brand of the switch."
    >>

    >My question is, has anyone seen a situation where the same broadcast
    >behavior occurs, but the CAM table itself is not overloaded and there is
    >no good reason for entries to be expiring? Furthermore, even if the
    >entries were expired, has anyone encountered situations (malicious or
    >otherwise), where a given port will receive traffic outside of its own
    >L2?
    >>

    >Thanks,
    >>

    >-jon


    Some router have an option of dumping all traffic to a give port, so
    if you are connected to the right router port you will see everything
    as if it was a hub. At least I already saw a router configured that
    way, that port that was connected to a computer that was dedicated to
    run snort.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.
  • No.6 | | 781 bytes | |

    Le vendredi 13 octobre 2006 * 09:32 -0700, Jon Hart a crit :

    Furthermore, even if the entries were expired, has anyone encountered
    situations (malicious or otherwise), where a given port will receive
    traffic outside of its own L2?

    I recently see that on a Cisco Catalyst 6500 L3 swith with an up to date
    IS image. some ports of a Gigabit slot, we sometimes see unicast
    trafic of others VLAN or destinated to other IP addresses of the same
    VLAN.

    I wasn't able to explain this behaviour, it could be a hardware problem.

    Nicob

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.7 | | 3257 bytes | |

    Sorry for my simple question.
    But what kind of broadcast do you discover?
    Perhaps you have a loop in your network, or even a sort of spanning tree
    (double connected wire to switch) which spams your network?!

    Florian

    Ben Nell wrote:
    Is all traffic being "broadcasted"? Can you narrow it down to a
    specific host that's common to all of the traffic, perhaps a gateway
    device? If you're doing multicasting on a gateway device
    (multicasting using unicast addressing), you would get the type of
    behavior that you're describing. I've seen this exact situation
    before, actually.

    BN

    10/13/06, Jon Hart <jhart (AT) spoofed (DOT) orgwrote:
    >Greetings,
    >>

    >I've got a situation here that I can't quite figure out. It is well
    >known that it is possible to cause a switched network to act like an
    >unswitched network by flooding the CAM table. There are countless tools
    >and documents out there that cover the offensive and defensive measures
    >related to this issue.
    >>

    >While this isn't Cisco's official documentation on this issue,
    >http://xrl.us/r8k7 says:
    >>

    >"Content-addressable memory (CAM) overflow: A CAM table is used to
    >determine where to direct incoming frames depending on which port the
    >incoming MAC address came from. When the CAM receives a frame with an
    >unknown destination, the proper procedure is to flood frames within
    >the acceptable Layer 2 domain (the proper VLAN). Hardware and
    >software tools are available (some for free), that can flood a switch
    >with MAC addresses. the CAM table limit is exceeded, switches
    >behave differently depending on the brand of the switch."
    >>

    >My question is, has anyone seen a situation where the same broadcast
    >behavior occurs, but the CAM table itself is not overloaded and there is
    >no good reason for entries to be expiring? Furthermore, even if the
    >entries were expired, has anyone encountered situations (malicious or
    >otherwise), where a given port will receive traffic outside of its own
    >L2?
    >>

    >Thanks,
    >>

    >-jon
    >>
    >>

    >
    >This List Sponsored by: Cenzic
    >>

    >Need to secure your web apps?
    >Cenzic Hailstorm finds vulnerabilities fast.
    >Click the link to buy it, try it or download Hailstorm for FREE.
    >
    >>

    >
    >>
    >>


    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.8 | | 2868 bytes | |

    This is a common behaviour with M$ NLB. X-S
    Analogously (e.g.: for load balancing) it could be possible that one of
    the end-point of the communications you're seeing is using with ARP
    replies that contain a MAC address that doesn't match the one used in
    it's ethernet frames (i.e.: the one that switches learn).

    dom, 2006-10-15 at 20:03 -0500, Ben Nell wrote:
    Is all traffic being "broadcasted"? Can you narrow it down to a
    specific host that's common to all of the traffic, perhaps a gateway
    device? If you're doing multicasting on a gateway device
    (multicasting using unicast addressing), you would get the type of
    behavior that you're describing. I've seen this exact situation
    before, actually.

    BN

    10/13/06, Jon Hart <jhart (AT) spoofed (DOT) orgwrote:
    Greetings,

    I've got a situation here that I can't quite figure out. It is well
    known that it is possible to cause a switched network to act like an
    unswitched network by flooding the CAM table. There are countless tools
    and documents out there that cover the offensive and defensive measures
    related to this issue.

    While this isn't Cisco's official documentation on this issue,
    http://xrl.us/r8k7 says:

    "Content-addressable memory (CAM) overflow: A CAM table is used to
    determine where to direct incoming frames depending on which port the
    incoming MAC address came from. When the CAM receives a frame with an
    unknown destination, the proper procedure is to flood frames within
    the acceptable Layer 2 domain (the proper VLAN). Hardware and
    software tools are available (some for free), that can flood a switch
    with MAC addresses. the CAM table limit is exceeded, switches
    behave differently depending on the brand of the switch."

    My question is, has anyone seen a situation where the same broadcast
    behavior occurs, but the CAM table itself is not overloaded and there is
    no good reason for entries to be expiring? Furthermore, even if the
    entries were expired, has anyone encountered situations (malicious or
    otherwise), where a given port will receive traffic outside of its own
    L2?

    Thanks,

    -jon
    --

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    --

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.9 | | 1314 bytes | |

    This can be done both on switches and routers. Cisco routers you would use a route-map to basically copy all traffic passing through one interface to another interface. Possible uses would be for IDS setups and probably other uses. switches it would be known as port mirroring.

    David Swafford.

    Ron <ron (AT) gwndev (DOT) com10/16/2006 3:49 pm

    Some router have an option of dumping all traffic to a give port, so
    if you are connected to the right router port you will see everything
    as if it was a hub. At least I already saw a router configured that
    way, that port that was connected to a computer that was dedicated to
    run snort.
    Just to clarify, I'm pretty sure you're talking about switches that have
    a "mirror" port.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

    Founded in Faith - Preserved with Pride - Sustained by Spirit

    Upcoming Events:
    ALTER PEN HUSE
    November 16
    7 - 9 p.m.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.10 | | 905 bytes | |

    Mon, 16, 2006 at 03:55:43PM -0400, Buz Dale wrote:
    I can think if a couple of possibilities. 1) This is
    broadcast/multicast traffic. 2) The mac addresses are unknown to the
    switch (So it will flood to find them.) 3) The port could be a trunk or
    a mirror of a trunk.

    I am also seeing normal broadcast/multicast traffic, but that is to be
    expected. #3 is not the case here.

    As for #2, thats kinda where I was going with my original question --
    why would a switch that is processing a session between two endpoints
    suddently forget the MAC? Yes, there are timeouts in play here, but
    aren't those along the lines of several minutes?

    Thanks,
    -jon

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.11 | | 980 bytes | |

    Hello Jon,

    As for #2, thats kinda where I was going with my original question --
    why would a switch that is processing a session between two endpoints
    suddently forget the MAC? Yes, there are timeouts in play here, but
    aren't those along the lines of several minutes?

    I'm no switch expert, but your last comment caused an hypothesis to pop
    into mind. Do any of your hosts have hard-coded MAC addresses set up?
    This isn't common, but if you're trying to prevent ARP poisoning, one
    might do this. If you were to do this, and not tell the switch which
    ports had those MACs, then it wouldn't get a chance to learn those MACs
    since those hosts wouldn't bother sending ARP requests, right? Just a
    thought.

    This List Sponsored by: Cenzic

    Need to secure your web apps?
    Cenzic Hailstorm finds vulnerabilities fast.
    Click the link to buy it, try it or download Hailstorm for FREE.

  • No.12 | | 388 bytes | |

    All,

    I've let the last few posts on this subject today go through (you'll be
    seeing them hit your inbox shortly) but unless this steers back toward a
    pen-test focused discussion I'll reject further posts. The topic is
    interesting and has covered a lot of routing concepts and aspects but this
    is a pen-testing list and not Cisco support :)

    Thanks,
  • No.13 | | 575 bytes | |

    Usually a lurker - try looking for macof for CAM overflow attacks and
    think basic traffic flooding.

    Found this as a pretty good start

    -Dave

    Erin Carroll wrote:
    All,

    I've let the last few posts on this subject today go through (you'll be
    seeing them hit your inbox shortly) but unless this steers back toward a
    pen-test focused discussion I'll reject further posts. The topic is
    interesting and has covered a lot of routing concepts and aspects but this
    is a pen-testing list and not Cisco support :)

    Thanks,

Re: unswitched behavior of a switched network...


max 4000 letters.
Your nickname that display:
In order to stop the spam: 8 + 7 =
QUESTION ON "Security"

EMSDN.COM